From: Subject: The Anatomy of a Search Engine Date: Wed, 9 Jan 2008 07:51:07 +0100 MIME-Version: 1.0 Content-Type: multipart/related; type="text/html"; boundary="----=_NextPart_000_0000_01C85294.643E2190" X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.3198 This is a multi-part message in MIME format. ------=_NextPart_000_0000_01C85294.643E2190 Content-Type: text/html; charset="utf-8" Content-Transfer-Encoding: quoted-printable Content-Location: http://infolab.stanford.edu/~backrub/google.html =EF=BB=BF The Anatomy of a Search Engine

The Anatomy of a Large-Scale Hypertextual Web Search = Engine

Sergey Brin and Lawrence Page {sergey, page}@cs.stanford.edu Computer Science Department, Stanford University, Stanford, CA=20 94305

Abstract

       In this paper, we = present=20 Google, a prototype of a large-scale search engine which makes heavy = use of=20 the structure present in hypertext. Google is designed to crawl and = index the=20 Web efficiently and produce much more satisfying search results than = existing=20 systems. The prototype with a full text and hyperlink database of at = least 24=20 million pages is available at http://google.stanford.edu/=20
       To engineer a search engine = is a=20 challenging task. Search engines index tens to hundreds of millions of = web=20 pages involving a comparable number of distinct terms. They answer = tens of=20 millions of queries every day. Despite the importance of large-scale = search=20 engines on the web, very little academic research has been done on = them.=20 Furthermore, due to rapid advance in technology and web proliferation, = creating a web search engine today is very different from three years = ago.=20 This paper provides an in-depth description of our large-scale web = search=20 engine -- the first such detailed public description we know of to = date.=20
       Apart from the problems of = scaling=20 traditional search techniques to data of this magnitude, there are new = technical challenges involved with using the additional information = present in=20 hypertext to produce better search results. This paper addresses this = question=20 of how to build a practical large-scale system which can exploit the=20 additional information present in hypertext. Also we look at the = problem of=20 how to effectively deal with uncontrolled hypertext collections where = anyone=20 can publish anything they want.=20 Keywords: World Wide Web, Search Engines, = Information=20 Retrieval, PageRank, Google

1. Introduction

(Note: There are two versions of this paper = -- a=20 longer full version and a shorter printed version. The full version is = available=20 on the web and the conference CD-ROM.)
The web creates new = challenges=20 for information retrieval. The amount of information on the web is = growing=20 rapidly, as well as the number of new users inexperienced in the art of = web=20 research. People are likely to surf the web using its link graph, often = starting=20 with high quality human maintained indices such as Yahoo! or with search engines. Human = maintained=20 lists cover popular topics effectively but are subjective, expensive to = build=20 and maintain, slow to improve, and cannot cover all esoteric topics. = Automated=20 search engines that rely on keyword matching usually return too many low = quality=20 matches. To make matters worse, some advertisers attempt to gain = people's=20 attention by taking measures meant to mislead automated search engines. = We have=20 built a large-scale search engine which addresses many of the problems = of=20 existing systems. It makes especially heavy use of the additional = structure=20 present in hypertext to provide much higher quality search results. We = chose our=20 system name, Google, because it is a common spelling of googol, or=20 10¹⁰⁰ and fits well with our goal of building very = large-scale search=20 engines.=20

1.1 Web Search Engines -- Scaling Up: 1994 - 2000

Search engine=20 technology has had to scale dramatically to keep up with the growth of = the web.=20 In 1994, one of the first web search engines, the World Wide Web Worm = (WWWW) [McBry= an 94]=20 had an index of 110,000 web pages and web accessible documents. As = of=20 November, 1997, the top search engines claim to index from 2 million=20 (WebCrawler) to 100 million web documents (from Search Engine Watch). It = is=20 foreseeable that by the year 2000, a comprehensive index of the Web will = contain=20 over a billion documents. At the same time, the number of queries search = engines=20 handle has grown incredibly too. In March and April 1994, the World Wide = Web=20 Worm received an average of about 1500 queries per day. In November = 1997,=20 Altavista claimed it handled roughly 20 million queries per day. With = the=20 increasing number of users on the web, and automated systems which query = search=20 engines, it is likely that top search engines will handle hundreds of = millions=20 of queries per day by the year 2000. The goal of our system is to = address many=20 of the problems, both in quality and scalability, introduced by scaling = search=20 engine technology to such extraordinary numbers.=20

1.2. Google: Scaling with the Web

Creating a search engine which = scales=20 even to today's web presents many challenges. Fast crawling technology = is needed=20 to gather the web documents and keep them up to date. Storage space must = be used=20 efficiently to store indices and, optionally, the documents themselves. = The=20 indexing system must process hundreds of gigabytes of data efficiently. = Queries=20 must be handled quickly, at a rate of hundreds to thousands per second.=20

These tasks are becoming increasingly difficult as the Web grows. = However,=20 hardware performance and cost have improved dramatically to partially = offset the=20 difficulty. There are, however, several notable exceptions to this = progress such=20 as disk seek time and operating system robustness. In designing Google, = we have=20 considered both the rate of growth of the Web and technological changes. = Google=20 is designed to scale well to extremely large data sets. It makes = efficient use=20 of storage space to store the index. Its data structures are optimized = for fast=20 and efficient access (see section 4.2). = Further,=20 we expect that the cost to index and store text or HTML will eventually = decline=20 relative to the amount that will be available (see Appendix = B). This=20 will result in favorable scaling properties for centralized systems like = Google.=20

1.3 Design Goals

1.3.1 Improved Search Quality

Our main goal is to improve the = quality of=20 web search engines. In 1994, some people believed that a complete search = index=20 would make it possible to find anything easily. According to Best of the Web = 1994 --=20 Navigators, "The best navigation service should make it easy = to find=20 almost anything on the Web (once all the data is entered)." = However, the=20 Web of 1997 is quite different. Anyone who has used a search engine = recently,=20 can readily testify that the completeness of the index is not the only = factor in=20 the quality of search results. "Junk results" often wash out any results = that a=20 user is interested in. In fact, as of November 1997, only one of the top = four=20 commercial search engines finds itself (returns its own search page in = response=20 to its name in the top ten results). One of the main causes of this = problem is=20 that the number of documents in the indices has been increasing by many = orders=20 of magnitude, but the user's ability to look at documents has not. = People are=20 still only willing to look at the first few tens of results. Because of = this, as=20 the collection size grows, we need tools that have very high precision = (number=20 of relevant documents returned, say in the top tens of results). Indeed, = we want=20 our notion of "relevant" to only include the very best documents since = there may=20 be tens of thousands of slightly relevant documents. This very high = precision is=20 important even at the expense of recall (the total number of relevant = documents=20 the system is able to return). There is quite a bit of recent optimism = that the=20 use of more hypertextual information can help improve search and other=20 applications [Marchiori = 97]=20 [Spertus = 97]=20 [Weiss = 96]=20 [Kleinberg=20 98]. In particular, link structure [Page = 98] and=20 link text provide a lot of information for making relevance judgments = and=20 quality filtering. Google makes use of both link structure and anchor = text (see=20 Sections 2.1=20 and 2.2)= .=20

1.3.2 Academic Search Engine Research

Aside from tremendous = growth, the=20 Web has also become increasingly commercial over time. In 1993, 1.5% of = web=20 servers were on .com domains. This number grew to over 60% in 1997. At = the same=20 time, search engines have migrated from the academic domain to the = commercial.=20 Up until now most search engine development has gone on at companies = with little=20 publication of technical details. This causes search engine technology = to remain=20 largely a black art and to be advertising oriented (see Appendix = A). With=20 Google, we have a strong goal to push more development and understanding = into=20 the academic realm.=20

Another important design goal was to build systems that reasonable = numbers of=20 people can actually use. Usage was important to us because we think some = of the=20 most interesting research will involve leveraging the vast amount of = usage data=20 that is available from modern web systems. For example, there are many = tens of=20 millions of searches performed every day. However, it is very difficult = to get=20 this data, mainly because it is considered commercially valuable.=20

Our final design goal was to build an architecture that can support = novel=20 research activities on large-scale web data. To support novel research = uses,=20 Google stores all of the actual documents it crawls in compressed form. = One of=20 our main goals in designing Google was to set up an environment where = other=20 researchers can come in quickly, process large chunks of the web, and = produce=20 interesting results that would have been very difficult to produce = otherwise. In=20 the short time the system has been up, there have already been several = papers=20 using databases generated by Google, and many others are underway. = Another goal=20 we have is to set up a Spacelab-like environment where researchers or = even=20 students can propose and do interesting experiments on our large-scale = web data.=20

2. System Features

The Google search engine has two important = features=20 that help it produce high precision results. First, it makes use of the = link=20 structure of the Web to calculate a quality ranking for each web page. = This=20 ranking is called PageRank and is described in detail in [Page 98]. = Second,=20 Google utilizes link to improve search results.=20

2.1 PageRank: Bringing Order to the Web

The = citation=20 (link) graph of the web is an important resource that has largely gone = unused in=20 existing web search engines. We have created maps containing as many as = 518=20 million of these hyperlinks, a significant sample of the total. These = maps allow=20 rapid calculation of a web page's "PageRank", an objective measure of = its=20 citation importance that corresponds well with people's subjective idea = of=20 importance. Because of this correspondence, PageRank is an excellent way = to=20 prioritize the results of web keyword searches. For most popular = subjects, a=20 simple text matching search that is restricted to web page titles = performs=20 admirably when PageRank prioritizes the results (demo available at google.stanford.edu). For the = type of=20 full text searches in the main Google system, PageRank also helps a = great deal.=20

2.1.1 Description of PageRank Calculation

Academic citation = literature=20 has been applied to the web, largely by counting citations or backlinks = to a=20 given page. This gives some approximation of a page's importance or = quality.=20 PageRank extends this idea by not counting links from all pages equally, = and by=20 normalizing by the number of links on a page. PageRank is defined as = follows:=20

We assume page A has pages T1...Tn which point to it = (i.e., are=20 citations). The parameter d is a damping factor which can be set = between 0 and=20 1. We usually set d to 0.85. There are more details about d in the = next=20 section. Also C(A) is defined as the number of links going out of page = A. The=20 PageRank of a page A is given as follows:=20
PR(A) =3D (1-d) + d (PR(T1)/C(T1) + ... + PR(Tn)/C(Tn))=20
Note that the PageRanks form a probability distribution over web = pages,=20 so the sum of all web pages' PageRanks will be = one.

PageRank=20 or PR(A) can be calculated using a simple iterative algorithm, = and=20 corresponds to the principal eigenvector of the normalized link matrix = of the=20 web. Also, a PageRank for 26 million web pages can be computed in a few = hours on=20 a medium size workstation. There are many other details which are beyond = the=20 scope of this paper.=20

2.1.2 Intuitive Justification

PageRank can be thought of as a = model of=20 user behavior. We assume there is a "random surfer" who is given a web = page at=20 random and keeps clicking on links, never hitting "back" but eventually = gets=20 bored and starts on another random page. The probability that the random = surfer=20 visits a page is its PageRank. And, the d damping factor is the=20 probability at each page the "random surfer" will get bored and request = another=20 random page. One important variation is to only add the damping factor = d=20 to a single page, or a group of pages. This allows for personalization = and can=20 make it nearly impossible to deliberately mislead the system in order to = get a=20 higher ranking. We have several other extensions to PageRank, again see = [Page = 98].=20

Another intuitive justification is that a page can have a high = PageRank if=20 there are many pages that point to it, or if there are some pages that = point to=20 it and have a high PageRank. Intuitively, pages that are well cited from = many=20 places around the web are worth looking at. Also, pages that have = perhaps only=20 one citation from something like the Yahoo!=20 homepage are also generally worth looking at. If a page was not high = quality, or=20 was a broken link, it is quite likely that Yahoo's homepage would not = link to=20 it. PageRank handles both these cases and everything in between by = recursively=20 propagating weights through the link structure of the web.=20

2.2 Anchor Text

The text of links is = treated in a=20 special way in our search engine. Most search engines associate the text = of a=20 link with the page that the link is on. In addition, we associate it = with the=20 page the link points to. This has several advantages. First, anchors = often=20 provide more accurate descriptions of web pages than the pages = themselves.=20 Second, anchors may exist for documents which cannot be indexed by a = text-based=20 search engine, such as images, programs, and databases. This makes it = possible=20 to return web pages which have not actually been crawled. Note that = pages that=20 have not been crawled can cause problems, since they are never checked = for=20 validity before being returned to the user. In this case, the search = engine can=20 even return a page that never actually existed, but had hyperlinks = pointing to=20 it. However, it is possible to sort the results, so that this particular = problem=20 rarely happens.=20

This idea of propagating anchor text to the page it refers to was = implemented=20 in the World Wide Web Worm [McBryan = 94]=20 especially because it helps search non-text information, and expands the = search=20 coverage with fewer downloaded documents. We use anchor propagation = mostly=20 because anchor text can help provide better quality results. Using = anchor text=20 efficiently is technically difficult because of the large amounts of = data which=20 must be processed. In our current crawl of 24 million pages, we had over = 259=20 million anchors which we indexed.=20

2.3 Other Features

Aside from PageRank and the use of anchor = text,=20 Google has several other features. First, it has location information = for all=20 hits and so it makes extensive use of proximity in search. Second, = Google keeps=20 track of some visual presentation details such as font size of words. = Words in a=20 larger or bolder font are weighted higher than other words. Third, full = raw HTML=20 of pages is available in a repository.=20

3 Related Work

Search research on the web has a short and = concise=20 history. The World Wide Web Worm (WWWW) [McBry= an 94]=20 was one of the first web search engines. It was subsequently = followed by=20 several other academic search engines, many of which are now public = companies.=20 Compared to the growth of the Web and the importance of search engines = there are=20 precious few documents about recent search engines [Pinkerton 94]. = According to=20 Michael Mauldin (chief scientist, Lycos Inc) [Mauldin],=20 "the various services (including Lycos) closely guard the details of = these=20 databases". However, there has been a fair amount of work on specific = features=20 of search engines. Especially well represented is work which can get = results by=20 post-processing the results of existing commercial search engines, or = produce=20 small scale "individualized" search engines. Finally, there has been a = lot of=20 research on information retrieval systems, especially on well controlled = collections. In the next two sections, we discuss some areas where this = research=20 needs to be extended to work better on the web.=20

3.1 Information Retrieval

Work in information retrieval systems = goes=20 back many years and is well developed [Witten = 94].=20 However, most of the research on information retrieval systems is on = small well=20 controlled homogeneous collections such as collections of scientific = papers or=20 news stories on a related topic. Indeed, the primary benchmark for = information=20 retrieval, the Text Retrieval Conference [TREC = 96], uses a=20 fairly small, well controlled collection for their benchmarks. The "Very = Large=20 Corpus" benchmark is only 20GB compared to the 147GB from our crawl of = 24=20 million web pages. Things that work well on TREC often do not produce = good=20 results on the web. For example, the standard vector space model tries = to return=20 the document that most closely approximates the query, given that both = query and=20 document are vectors defined by their word occurrence. On the web, this = strategy=20 often returns very short documents that are the query plus a few words. = For=20 example, we have seen a major search engine return a page containing = only "Bill=20 Clinton Sucks" and picture from a "Bill Clinton" query. Some argue that = on the=20 web, users should specify more accurately what they want and add more = words to=20 their query. We disagree vehemently with this position. If a user issues = a query=20 like "Bill Clinton" they should get reasonable results since there is a = enormous=20 amount of high quality information available on this topic. Given = examples like=20 these, we believe that the standard information retrieval work needs to = be=20 extended to deal effectively with the web.=20

3.2 Differences Between the Web and Well Controlled = Collections

The web=20 is a vast collection of completely uncontrolled heterogeneous documents. = Documents on the web have extreme variation internal to the documents, = and also=20 in the external meta information that might be available. For example, = documents=20 differ internally in their language (both human and programming), = vocabulary=20 (email addresses, links, zip codes, phone numbers, product numbers), = type or=20 format (text, HTML, PDF, images, sounds), and may even be machine = generated (log=20 files or output from a database). On the other hand, we define external = meta=20 information as information that can be inferred about a document, but is = not=20 contained within it. Examples of external meta information include = things like=20 reputation of the source, update frequency, quality, popularity or = usage, and=20 citations. Not only are the possible sources of external meta = information=20 varied, but the things that are being measured vary many orders of = magnitude as=20 well. For example, compare the usage information from a major homepage, = like=20 Yahoo's which currently receives millions of page views every day with = an=20 obscure historical article which might receive one view every ten years. = Clearly, these two items must be treated very differently by a search = engine.=20

Another big difference between the web and traditional well = controlled=20 collections is that there is virtually no control over what people can = put on=20 the web. Couple this flexibility to publish anything with the enormous = influence=20 of search engines to route traffic and companies which deliberately = manipulating=20 search engines for profit become a serious problem. This problem that = has not=20 been addressed in traditional closed information retrieval systems. = Also, it is=20 interesting to note that metadata efforts have largely failed with web = search=20 engines, because any text on the page which is not directly represented = to the=20 user is abused to manipulate search engines. There are even numerous = companies=20 which specialize in manipulating search engines for profit.=20

4 System Anatomy

First, we will provide a high level discussion = of the=20 architecture. Then, there is some in-depth descriptions of important = data=20 structures. Finally, the major applications: crawling, indexing, and = searching=20 will be examined in depth.=20

Figure 1. High Level Google=20 = Architecture

&nbs= p;=20

4.1 Google Architecture Overview

In this section, we will give a = high=20 level overview of how the whole system works as pictured in Figure 1. = Further=20 sections will discuss the applications and data structures not mentioned = in this=20 section. Most of Google is implemented in C or C++ for efficiency and = can run in=20 either Solaris or Linux.=20

In Google, the web crawling (downloading of web pages) is done by = several=20 distributed crawlers. There is a URLserver that sends lists of URLs to = be=20 fetched to the crawlers. The web pages that are fetched are then sent to = the=20 storeserver. The storeserver then compresses and stores the web pages = into a=20 repository. Every web page has an associated ID number called a docID = which is=20 assigned whenever a new URL is parsed out of a web page. The indexing = function=20 is performed by the indexer and the sorter. The indexer performs a = number of=20 functions. It reads the repository, uncompresses the documents, and = parses them.=20 Each document is converted into a set of word occurrences called hits. = The hits=20 record the word, position in document, an approximation of font size, = and=20 capitalization. The indexer distributes these hits into a set of = "barrels",=20 creating a partially sorted forward index. The indexer performs another=20 important function. It parses out all the links in every web page and = stores=20 important information about them in an anchors file. This file contains = enough=20 information to determine where each link points from and to, and the = text of the=20 link.=20

The URLresolver reads the anchors file and converts relative URLs = into=20 absolute URLs and in turn into docIDs. It puts the anchor text into the = forward=20 index, associated with the docID that the anchor points to. It also = generates a=20 database of links which are pairs of docIDs. The links database is used = to=20 compute PageRanks for all the documents.=20

The sorter takes the barrels, which are sorted by docID (this is a=20 simplification, see Section = 4.2.5),=20 and resorts them by wordID to generate the inverted index. This is done = in place=20 so that little temporary space is needed for this operation. The sorter = also=20 produces a list of wordIDs and offsets into the inverted index. A = program called=20 DumpLexicon takes this list together with the lexicon produced by the = indexer=20 and generates a new lexicon to be used by the searcher. The searcher is = run by a=20 web server and uses the lexicon built by DumpLexicon together with the = inverted=20 index and the PageRanks to answer queries.=20

4.2 Major Data Structures

Google's data = structures are=20 optimized so that a large document collection can be crawled, indexed, = and=20 searched with little cost. Although, CPUs and bulk input output rates = have=20 improved dramatically over the years, a disk seek still requires about = 10 ms to=20 complete. Google is designed to avoid disk seeks whenever possible, and = this has=20 had a considerable influence on the design of the data structures.=20

4.2.1 BigFiles

BigFiles are virtual files spanning multiple file = systems=20 and are addressable by 64 bit integers. The allocation among multiple = file=20 systems is handled automatically. The BigFiles package also handles = allocation=20 and deallocation of file descriptors, since the operating systems do not = provide=20 enough for our needs. BigFiles also support rudimentary compression = options.=20

4.2.2 Repository

=20

Figure 2. Repository Data=20 Structure

The = repository=20 contains the full HTML of every web page. Each page is compressed using = zlib=20 (see RFC= 1950).=20 The choice of compression technique is a tradeoff between speed and = compression=20 ratio. We chose zlib's speed over a significant improvement in = compression=20 offered by bzip. The = compression=20 rate of bzip was approximately 4 to 1 on the repository as compared to = zlib's 3=20 to 1 compression. In the repository, the documents are stored one after = the=20 other and are prefixed by docID, length, and URL as can be seen in = Figure 2. The=20 repository requires no other data structures to be used in order to = access it.=20 This helps with data consistency and makes development much easier; we = can=20 rebuild all the other data structures from only the repository and a = file which=20 lists crawler errors.=20

4.2.3 Document Index

The document index keeps information about = each=20 document. It is a fixed width ISAM (Index sequential access mode) index, = ordered=20 by docID. The information stored in each entry includes the current = document=20 status, a pointer into the repository, a document checksum, and various=20 statistics. If the document has been crawled, it also contains a pointer = into a=20 variable width file called docinfo which contains its URL and title. = Otherwise=20 the pointer points into the URLlist which contains just the URL. This = design=20 decision was driven by the desire to have a reasonably compact data = structure,=20 and the ability to fetch a record in one disk seek during a search=20

Additionally, there is a file which is used to convert URLs into = docIDs. It=20 is a list of URL checksums with their corresponding docIDs and is sorted = by=20 checksum. In order to find the docID of a particular URL, the URL's = checksum is=20 computed and a binary search is performed on the checksums file to find = its=20 docID. URLs may be converted into docIDs in batch by doing a merge with = this=20 file. This is the technique the URLresolver uses to turn URLs into = docIDs. This=20 batch mode of update is crucial because otherwise we must perform one = seek for=20 every link which assuming one disk would take more than a month for our = 322=20 million link dataset.=20

4.2.4 Lexicon

The lexicon has several different forms. One = important=20 change from earlier systems is that the lexicon can fit in memory for a=20 reasonable price. In the current implementation we can keep the lexicon = in=20 memory on a machine with 256 MB of main memory. The current lexicon = contains 14=20 million words (though some rare words were not added to the lexicon). It = is=20 implemented in two parts -- a list of the words (concatenated together = but=20 separated by nulls) and a hash table of pointers. For various functions, = the=20 list of words has some auxiliary information which is beyond the scope = of this=20 paper to explain fully.=20

4.2.5 Hit Lists

A hit list corresponds to a = list of=20 occurrences of a particular word in a particular document including = position,=20 font, and capitalization information. Hit lists account for most of the = space=20 used in both the forward and the inverted indices. Because of this, it = is=20 important to represent them as efficiently as possible. We considered = several=20 alternatives for encoding position, font, and capitalization -- simple = encoding=20 (a triple of integers), a compact encoding (a hand optimized allocation = of=20 bits), and Huffman coding. In the end we chose a hand optimized compact = encoding=20 since it required far less space than the simple encoding and far less = bit=20 manipulation than Huffman coding. The details of the hits are shown in = Figure 3.=20

Our compact encoding uses two bytes for every hit. There are two = types of=20 hits: fancy hits and plain hits. Fancy hits include hits occurring in a = URL,=20 title, anchor text, or meta tag. Plain hits include everything else. A = plain hit=20 consists of a capitalization bit, font size, and 12 bits of word = position in a=20 document (all positions higher than 4095 are labeled 4096). Font size is = represented relative to the rest of the document using three bits (only = 7 values=20 are actually used because 111 is the flag that signals a fancy hit). A = fancy hit=20 consists of a capitalization bit, the font size set to 7 to indicate it = is a=20 fancy hit, 4 bits to encode the type of fancy hit, and 8 bits of = position. For=20 anchor hits, the 8 bits of position are split into 4 bits for position = in anchor=20 and 4 bits for a hash of the docID the anchor occurs in. This gives us = some=20 limited phrase searching as long as there are not that many anchors for = a=20 particular word. We expect to update the way that anchor hits are stored = to=20 allow for greater resolution in the position and docIDhash fields. We = use font=20 size relative to the rest of the document because when searching, you do = not=20 want to rank otherwise identical documents differently just because one = of the=20 documents is in a larger font.=20

Figure 3. Forward and Reverse Indexes and the=20 Lexicon

=20

The length of a hit list is stored before the hits themselves. To = save space,=20 the length of the hit list is combined with the wordID in the forward = index and=20 the docID in the inverted index. This limits it to 8 and 5 bits = respectively=20 (there are some tricks which allow 8 bits to be borrowed from the = wordID). If=20 the length is longer than would fit in that many bits, an escape code is = used in=20 those bits, and the next two bytes contain the actual length.=20

4.2.6 Forward Index

The forward index is actually already = partially=20 sorted. It is stored in a number of barrels (we used 64). Each barrel = holds a=20 range of wordID's. If a document contains words that fall into a = particular=20 barrel, the docID is recorded into the barrel, followed by a list of = wordID's=20 with hitlists which correspond to those words. This scheme requires = slightly=20 more storage because of duplicated docIDs but the difference is very = small for a=20 reasonable number of buckets and saves considerable time and coding = complexity=20 in the final indexing phase done by the sorter. Furthermore, instead of = storing=20 actual wordID's, we store each wordID as a relative difference from the = minimum=20 wordID that falls into the barrel the wordID is in. This way, we can use = just 24=20 bits for the wordID's in the unsorted barrels, leaving 8 bits for the = hit list=20 length.=20

4.2.7 Inverted Index

The inverted index consists of the same = barrels as=20 the forward index, except that they have been processed by the sorter. = For every=20 valid wordID, the lexicon contains a pointer into the barrel that wordID = falls=20 into. It points to a doclist of docID's together with their = corresponding hit=20 lists. This doclist represents all the occurrences of that word in all=20 documents.=20

An important issue is in what order the docID's should appear in the = doclist.=20 One simple solution is to store them sorted by docID. This allows for = quick=20 merging of different doclists for multiple word queries. Another option = is to=20 store them sorted by a ranking of the occurrence of the word in each = document.=20 This makes answering one word queries trivial and makes it likely that = the=20 answers to multiple word queries are near the start. However, merging is = much=20 more difficult. Also, this makes development much more difficult in that = a=20 change to the ranking function requires a rebuild of the index. We chose = a=20 compromise between these options, keeping two sets of inverted barrels = -- one=20 set for hit lists which include title or anchor hits and another set for = all hit=20 lists. This way, we check the first set of barrels first and if there = are not=20 enough matches within those barrels we check the larger ones.=20

4.3 Crawling the Web

Running a web crawler is a challenging = task. There=20 are tricky performance and reliability issues and even more importantly, = there=20 are social issues. Crawling is the most fragile application since it = involves=20 interacting with hundreds of thousands of web servers and various name = servers=20 which are all beyond the control of the system.=20

In order to scale to hundreds of millions of web pages, Google has a = fast=20 distributed crawling system. A single URLserver serves lists of URLs to = a number=20 of crawlers (we typically ran about 3). Both the URLserver and the = crawlers are=20 implemented in Python. Each crawler keeps roughly 300 connections open = at once.=20 This is necessary to retrieve web pages at a fast enough pace. At peak = speeds,=20 the system can crawl over 100 web pages per second using four crawlers. = This=20 amounts to roughly 600K per second of data. A major performance stress = is DNS=20 lookup. Each crawler maintains a its own DNS cache so it does not need = to do a=20 DNS lookup before crawling each document. Each of the hundreds of = connections=20 can be in a number of different states: looking up DNS, connecting to = host,=20 sending request, and receiving response. These factors make the crawler = a=20 complex component of the system. It uses asynchronous IO to manage = events, and a=20 number of queues to move page fetches from state to state.=20

It turns out that running a crawler which connects to more than half = a=20 million servers, and generates tens of millions of log entries generates = a fair=20 amount of email and phone calls. Because of the vast number of people = coming on=20 line, there are always those who do not know what a crawler is, because = this is=20 the first one they have seen. Almost daily, we receive an email = something like,=20 "Wow, you looked at a lot of pages from my web site. How did you like = it?" There=20 are also some people who do not know about the rob= ots=20 exclusion protocol, and think their page should be protected from = indexing=20 by a statement like, "This page is copyrighted and should not be = indexed", which=20 needless to say is difficult for web crawlers to understand. Also, = because of=20 the huge amount of data involved, unexpected things will happen. For = example,=20 our system tried to crawl an online game. This resulted in lots of = garbage=20 messages in the middle of their game! It turns out this was an easy = problem to=20 fix. But this problem had not come up until we had downloaded tens of = millions=20 of pages. Because of the immense variation in web pages and servers, it = is=20 virtually impossible to test a crawler without running it on large part = of the=20 Internet. Invariably, there are hundreds of obscure problems which may = only=20 occur on one page out of the whole web and cause the crawler to crash, = or worse,=20 cause unpredictable or incorrect behavior. Systems which access large = parts of=20 the Internet need to be designed to be very robust and carefully tested. = Since=20 large complex systems such as crawlers will invariably cause problems, = there=20 needs to be significant resources devoted to reading the email and = solving these=20 problems as they come up.=20

4.4 Indexing the Web

Parsing -- Any parser which is designed to run on the = entire Web=20 must handle a huge array of possible errors. These range from typos in = HTML=20 tags to kilobytes of zeros in the middle of a tag, non-ASCII = characters, HTML=20 tags nested hundreds deep, and a great variety of other errors that = challenge=20 anyone's imagination to come up with equally creative ones. For = maximum speed,=20 instead of using YACC to generate a CFG parser, we use flex to = generate a=20 lexical analyzer which we outfit with its own stack. Developing this = parser=20 which runs at a reasonable speed and is very robust involved a fair = amount of=20 work.=20
Indexing Documents into Barrels -- After each = document is=20 parsed, it is encoded into a number of barrels. Every word is = converted into a=20 wordID by using an in-memory hash table -- the lexicon. New additions = to the=20 lexicon hash table are logged to a file. Once the words are converted = into=20 wordID's, their occurrences in the current document are translated = into hit=20 lists and are written into the forward barrels. The main difficulty = with=20 parallelization of the indexing phase is that the lexicon needs to be = shared.=20 Instead of sharing the lexicon, we took the approach of writing a log = of all=20 the extra words that were not in a base lexicon, which we fixed at 14 = million=20 words. That way multiple indexers can run in parallel and then the = small log=20 file of extra words can be processed by one final indexer.=20
Sorting -- In order to generate the inverted index, the = sorter=20 takes each of the forward barrels and sorts it by wordID to produce an = inverted barrel for title and anchor hits and a full text inverted = barrel.=20 This process happens one barrel at a time, thus requiring little = temporary=20 storage. Also, we parallelize the sorting phase to use as many = machines as we=20 have simply by running multiple sorters, which can process different = buckets=20 at the same time. Since the barrels don't fit into main memory, the = sorter=20 further subdivides them into baskets which do fit into memory based on = wordID=20 and docID. Then the sorter, loads each basket into memory, sorts it = and writes=20 its contents into the short inverted barrel and the full inverted = barrel.=20

4.5 Searching

The goal of searching is to provide quality search = results=20 efficiently. Many of the large commercial search engines seemed to have = made=20 great progress in terms of efficiency. Therefore, we have focused more = on=20 quality of search in our research, although we believe our solutions are = scalable to commercial volumes with a bit more effort. The google query=20 evaluation process is show in Figure 4.=20

Parse the query.=20
Convert words into wordIDs.=20
Seek to the start of the doclist in the short barrel for = every word.=20
Scan through the doclists until there is a document that = matches all=20 the search terms.=20
Compute the rank of that document for the query.=20
If we are in the short barrels and at the end of any = doclist, seek=20 to the start of the doclist in the full barrel for every word = and go to=20 step 4.=20
If we are not at the end of any doclist go to step 4. =
Sort the=20 documents that have matched by rank and return the top = k.

Figure 4. Google Query=20 Evaluation

=20

To put a limit on response time, once a certain number (currently = 40,000) of=20 matching documents are found, the searcher automatically goes to step 8 = in=20 Figure 4. This means that it is possible that sub-optimal results would = be=20 returned. We are currently investigating other ways to solve this = problem. In=20 the past, we sorted the hits according to PageRank, which seemed to = improve the=20 situation.=20

4.5.1 The Ranking System

Google maintains much more information = about=20 web documents than typical search engines. Every hitlist includes = position,=20 font, and capitalization information. Additionally, we factor in hits = from=20 anchor text and the PageRank of the document. Combining all of this = information=20 into a rank is difficult. We designed our ranking function so that no = particular=20 factor can have too much influence. First, consider the simplest case -- = a=20 single word query. In order to rank a document with a single word query, = Google=20 looks at that document's hit list for that word. Google considers each = hit to be=20 one of several different types (title, anchor, URL, plain text large = font, plain=20 text small font, ...), each of which has its own type-weight. The = type-weights=20 make up a vector indexed by type. Google counts the number of hits of = each type=20 in the hit list. Then every count is converted into a count-weight.=20 Count-weights increase linearly with counts at first but quickly taper = off so=20 that more than a certain count will not help. We take the dot product of = the=20 vector of count-weights with the vector of type-weights to compute an IR = score=20 for the document. Finally, the IR score is combined with PageRank to = give a=20 final rank to the document.=20

For a multi-word search, the situation is more complicated. Now = multiple hit=20 lists must be scanned through at once so that hits occurring close = together in a=20 document are weighted higher than hits occurring far apart. The hits = from the=20 multiple hit lists are matched up so that nearby hits are matched = together. For=20 every matched set of hits, a proximity is computed. The proximity is = based on=20 how far apart the hits are in the document (or anchor) but is classified = into 10=20 different value "bins" ranging from a phrase match to "not even close". = Counts=20 are computed not only for every type of hit but for every type and = proximity.=20 Every type and proximity pair has a type-prox-weight. The counts are = converted=20 into count-weights and we take the dot product of the count-weights and = the=20 type-prox-weights to compute an IR score. All of these numbers and = matrices can=20 all be displayed with the search results using a special debug mode. = These=20 displays have been very helpful in developing the ranking system.=20

4.5.2 Feedback

The ranking function has many parameters like the = type-weights and the type-prox-weights. Figuring out the right values = for these=20 parameters is something of a black art. In order to do this, we have a = user=20 feedback mechanism in the search engine. A trusted user may optionally = evaluate=20 all of the results that are returned. This feedback is saved. Then when = we=20 modify the ranking function, we can see the impact of this change on all = previous searches which were ranked. Although far from perfect, this = gives us=20 some idea of how a change in the ranking function affects the search = results.=20

5 Results and Performance

=20

Query: bill clinton
http://www.whitehouse.gov/ <= /FONT> =20
100.00%

= (no date)=20 (0K)
http://www.whitehouse.gov/  =20
      Office of = the=20 President
        99.67%

= (Dec 23=20 1996) (2K)
       =20 http://www.whitehouse.gov/WH/EOP/OP/html/OP_Home.html =
      Welcome To The = White=20 House
        = 99.98% =20 (Nov 09 1997) (5K)
       =20 http://www.whitehouse.gov/WH/Welcome.html    =
      Send = Electronic Mail to the President
        = 99.86% =20 (Jul 14 1997) (5K)
       =20 = http://www.whitehouse.gov/WH/Mail/html/Mail_President.html &n= bsp;=20
mailto:president@whitehouse.gov<= /A>  =20
99.98%

      mailto:President@whitehouse.gov<= /A>  =20
       =20 99.27%

The "Unofficial" Bill=20 Clinton
94.06%

= (Nov 11=20 1997) (14K)
http://zpub.com/un/un-bc.html   =
       Bill Clinton Meets The=20 Shrinks
         = 86.27% =20 (Jun 29 1997) (63K)
        =20 http://zpub.com/un/un-bc9.html
President Bill = Clinton - The=20 Dark Side
97.27% =20 (Nov 10 1997) (15K)
http://www.realchange.org/clinton.htm   =
$3 Bill=20 Clinton
94.73%

= (no date)=20 (4K) = http://www.gatewy.net/~tjohnson/clinton1.html =20

Figure 4. Sample Results from=20 Google

The most = important=20 measure of a search engine is the quality of its search results. While a = complete user evaluation is beyond the scope of this paper, our own = experience=20 with Google has shown it to produce better results than the major = commercial=20 search engines for most searches. As an example which illustrates the = use of=20 PageRank, anchor text, and proximity, Figure 4 shows Google's results = for a=20 search on "bill clinton". These results demonstrates some of Google's = features.=20 The results are clustered by server. This helps considerably when = sifting=20 through result sets. A number of results are from the whitehouse.gov = domain=20 which is what one may reasonably expect from such a search. Currently, = most=20 major commercial search engines do not return any results from = whitehouse.gov,=20 much less the right ones. Notice that there is no title for the first = result.=20 This is because it was not crawled. Instead, Google relied on anchor = text to=20 determine this was a good answer to the query. Similarly, the fifth = result is an=20 email address which, of course, is not crawlable. It is also a result of = anchor=20 text.=20

All of the results are reasonably high quality pages and, at last = check, none=20 were broken links. This is largely because they all have high PageRank. = The=20 PageRanks are the percentages in red along with bar graphs. Finally, = there are=20 no results about a Bill other than Clinton or about a Clinton other than = Bill.=20 This is because we place heavy importance on the proximity of word = occurrences.=20 Of course a true test of the quality of a search engine would involve an = extensive user study or results analysis which we do not have room for = here.=20 Instead, we invite the reader to try Google for themselves at http://google.stanford.edu/.=20

5.1 Storage Requirements

Aside from search quality, Google is = designed=20 to scale cost effectively to the size of the Web as it grows. One aspect = of this=20 is to use storage efficiently. Table 1 has a breakdown of some = statistics and=20 storage requirements of Google. Due to compression the total size of the = repository is about 53 GB, just over one third of the total data it = stores. At=20 current disk prices this makes the repository a relatively cheap source = of=20 useful data. More importantly, the total of all the data used by the = search=20 engine requires a comparable amount of storage, about 55 GB. = Furthermore, most=20 queries can be answered using just the short inverted index. With better = encoding and compression of the Document Index, a high quality web = search engine=20 may fit onto a 7GB drive of a new PC.
=20

Storage Statistics
Total Size of Fetched Pages	147.8 GB
Compressed Repository	53.5 GB
Short Inverted Index	4.1 GB
Full Inverted Index	37.2 GB
Lexicon	293 MB
Temporary Anchor Data (not in total)	6.6 GB
Document Index Incl. Variable Width Data	9.7 GB
Links Database	3.9 GB
Total Without Repository	55.2 GB
Total With Repository	108.7 GB

Web Page Statistics
Number of Web Pages Fetched	24 million
Number of Urls Seen	76.5 million
Number of Email Addresses	1.7 million
Number of 404's	1.6 million

Table 1. = Statistics

=20

5.2 System Performance

It is important for a search engine = to=20 crawl and index efficiently. This way information can be kept up to date = and=20 major changes to the system can be tested relatively quickly. For = Google, the=20 major operations are Crawling, Indexing, and Sorting. It is difficult to = measure=20 how long crawling took overall because disks filled up, name servers = crashed, or=20 any number of other problems which stopped the system. In total it took = roughly=20 9 days to download the 26 million pages (including errors). However, = once the=20 system was running smoothly, it ran much faster, downloading the last 11 = million=20 pages in just 63 hours, averaging just over 4 million pages per day or = 48.5=20 pages per second. We ran the indexer and the crawler simultaneously. The = indexer=20 ran just faster than the crawlers. This is largely because we spent just = enough=20 time optimizing the indexer so that it would not be a bottleneck. These=20 optimizations included bulk updates to the document index and placement = of=20 critical data structures on the local disk. The indexer runs at roughly = 54 pages=20 per second. The sorters can be run completely in parallel; using four = machines,=20 the whole process of sorting takes about 24 hours.=20

5.3 Search Performance

Improving the performance of search was = not the=20 major focus of our research up to this point. The current version of = Google=20 answers most queries in between 1 and 10 seconds. This time is mostly = dominated=20 by disk IO over NFS (since disks are spread over a number of machines).=20 Furthermore, Google does not have any optimizations such as query = caching,=20 subindices on common terms, and other common optimizations. We intend to = speed=20 up Google considerably through distribution and hardware, software, and=20 algorithmic improvements. Our target is to be able to handle several = hundred=20 queries per second. Table 2 has some sample query times from the current = version=20 of Google. They are repeated to show the speedups resulting from cached = IO.=20

	Initial Query	Same Query Repeated (IO mostly=20 cached)
Query	CPU Time(s)	Total Time(s)	CPU Time(s)	Total Time(s)
al gore	0.09	2.13	0.06	0.06
vice president	1.77	3.84	1.66	1.80
hard disks	0.25	4.86	0.20	0.24
search engines	1.31	9.63	1.16	1.16

Table 2. Search=20 Times

=20

6 Conclusions

Google is designed to be a scalable search engine. = The=20 primary goal is to provide high quality search results over a rapidly = growing=20 World Wide Web. Google employs a number of techniques to improve search = quality=20 including page rank, anchor text, and proximity information. = Furthermore, Google=20 is a complete architecture for gathering web pages, indexing them, and=20 performing search queries over them.=20

6.1 Future Work

A large-scale web search engine is a complex = system and=20 much remains to be done. Our immediate goals are to improve search = efficiency=20 and to scale to approximately 100 million web pages. Some simple = improvements to=20 efficiency include query caching, smart disk allocation, and subindices. = Another=20 area which requires much research is updates. We must have smart = algorithms to=20 decide what old web pages should be recrawled and what new ones should = be=20 crawled. Work toward this goal has been done in [Cho = 98]. One=20 promising area of research is using proxy caches to build search = databases,=20 since they are demand driven. We are planning to add simple features = supported=20 by commercial search engines like boolean operators, negation, and = stemming.=20 However, other features are just starting to be explored such as = relevance=20 feedback and clustering (Google currently supports a simple hostname = based=20 clustering). We also plan to support user context (like the user's = location),=20 and result summarization. We are also working to extend the use of link=20 structure and link text. Simple experiments indicate PageRank can be=20 personalized by increasing the weight of a user's home page or = bookmarks. As for=20 link text, we are experimenting with using text surrounding links in = addition to=20 the link text itself. A Web search engine is a very rich environment for = research ideas. We have far too many to list here so we do not expect = this=20 Future Work section to become much shorter in the near future.=20

6.2 High Quality Search

The biggest problem facing users of web = search=20 engines today is the quality of the results they get back. While the = results are=20 often amusing and expand users' horizons, they are often frustrating and = consume=20 precious time. For example, the top result for a search for "Bill = Clinton" on=20 one of the most popular commercial search engines was the Bill Clinton = Joke of the=20 Day: April 14, 1997. Google is designed to provide higher quality = search so=20 as the Web continues to grow rapidly, information can be found easily. = In order=20 to accomplish this Google makes heavy use of hypertextual information = consisting=20 of link structure and link (anchor) text. Google also uses proximity and = font=20 information. While evaluation of a search engine is difficult, we have=20 subjectively found that Google returns higher quality search results = than=20 current commercial search engines. The analysis of link structure via = PageRank=20 allows Google to evaluate the quality of web pages. The use of link text = as a=20 description of what the link points to helps the search engine return = relevant=20 (and to some degree high quality) results. Finally, the use of proximity = information helps increase relevance a great deal for many queries.=20

6.3 Scalable Architecture

Aside from the quality of search, = Google is=20 designed to scale. It must be efficient in both space and time, and = constant=20 factors are very important when dealing with the entire Web. In = implementing=20 Google, we have seen bottlenecks in CPU, memory access, memory capacity, = disk=20 seeks, disk throughput, disk capacity, and network IO. Google has = evolved to=20 overcome a number of these bottlenecks during various operations. = Google's major=20 data structures make efficient use of available storage space. = Furthermore, the=20 crawling, indexing, and sorting operations are efficient enough to be = able to=20 build an index of a substantial portion of the web -- 24 million pages, = in less=20 than one week. We expect to be able to build an index of 100 million = pages in=20 less than a month.=20

6.4 A Research Tool

In addition to being a high quality search = engine,=20 Google is a research tool. The data Google has collected has already = resulted in=20 many other papers submitted to conferences and many more on the way. = Recent=20 research such as [Abiteboul = 97]=20 has shown a number of limitations to queries about the Web that may be = answered=20 without having the Web available locally. This means that Google (or a = similar=20 system) is not only a valuable research tool but a necessary one for a = wide=20 range of applications. We hope Google will be a resource for searchers = and=20 researchers all around the world and will spark the next generation of = search=20 engine technology.=20

7 Acknowledgments

Scott Hassan and Alan Steremberg have been = critical to=20 the development of Google. Their talented contributions are = irreplaceable, and=20 the authors owe them much gratitude. We would also like to thank Hector=20 Garcia-Molina, Rajeev Motwani, Jeff Ullman, and Terry Winograd and the = whole=20 WebBase group for their support and insightful discussions. Finally we = would=20 like to recognize the generous support of our equipment donors IBM, = Intel, and=20 Sun and our funders. The research described here was conducted as part = of the=20 Stanford Integrated Digital Library Project, supported by the National = Science=20 Foundation under Cooperative Agreement IRI-9411306. Funding for this = cooperative=20 agreement is also provided by DARPA and NASA, and by Interval Research, = and the=20 industrial partners of the Stanford Digital Libraries Project.=20

References

Best of the Web 1994 -- Navigators http://botw.org/1994= /awards/navigators.html=20
Bill Clinton Joke of the Day: April 14, 1997. http://www.io.com= /~cjburke/clinton/970414.html=20
Bzip2 Homepage http://www.muraroa.demon.co.uk/<= /A>=20
Google Search Engine http://google.stanford.edu/=20
Harvest http://harvest.transarc.com/ =
Mauldin, Michael L. Lycos Design Choices in an Internet Search = Service,=20 IEEE Expert Interview http://www.computer.org/pubs/expert/1997/trends/x1008/mauldin.htm=20
The Effect of Cellular Phone Use Upon Driver Attention http://www.we= bfirst.com/aaa/text/cell/cell0toc.htm=20
Search Engine Watch http://www.searchenginewatch.c= om/=20
RFC 1950 (zlib) ftp= ://ftp.uu.net/graphics/png/documents/zlib/zdoc-index.html=20
Robots Exclusion Protocol: ht= tp://info.webcrawler.com/mak/projects/robots/exclusion.html=20
Web Growth Summary: htt= p://www.mit.edu/people/mkgray/net/web-growth-summary.html=20
Yahoo! http://www.yahoo.com/

[Abiteboul 97] Serge Abiteboul and Victor Vianu, Queries and=20 Computation on the Web. Proceedings of the International = Conference on=20 Database Theory. Delphi, Greece 1997.=20
[Bagdikian 97] Ben H. Bagdikian. The Media Monopoly. 5th = Edition.=20 Publisher: Beacon, ISBN: 0807061557=20
[Chakrabarti 98] S.Chakrabarti, B.Dom, D.Gibson, J.Kleinberg, P. = Raghavan=20 and S. Rajagopalan. Automatic Resource Compilation by Analyzing = Hyperlink=20 Structure and Associated Text. Seventh International Web = Conference (WWW=20 98). Brisbane, Australia, April 14-18, 1998.=20
[Cho 98] Junghoo Cho, Hector Garcia-Molina, Lawrence Page. = Efficient=20 Crawling Through URL Ordering. Seventh International Web = Conference (WWW=20 98). Brisbane, Australia, April 14-18, 1998.=20
[Gravano 94] Luis Gravano, Hector Garcia-Molina, and A. Tomasic. = The=20 Effectiveness of GlOSS for the Text-Database Discovery Problem. = Proc. of=20 the 1994 ACM SIGMOD International Conference On Management Of Data, = 1994.=20
[Kleinberg 98] Jon Kleinberg, Authoritative Sources in a = Hyperlinked=20 Environment, Proc. ACM-SIAM Symposium on Discrete Algorithms, = 1998.=20
[Marchiori 97] Massimo Marchiori. The Quest for Correct = Information on=20 the Web: Hyper Search Engines. The Sixth International WWW = Conference (WWW=20 97). Santa Clara, USA, April 7-11, 1997.=20
[McBryan 94] Oliver A. McBryan. GENVL and WWWW: Tools for = Taming the=20 Web. First International Conference on the World Wide Web. CERN, = Geneva=20 (Switzerland), May 25-26-27 1994. http:/= /www.cs.colorado.edu/home/mcbryan/mypapers/www94.ps=20
[Page 98] Lawrence Page, Sergey Brin, Rajeev Motwani, Terry = Winograd.=20 The PageRank Citation Ranking: Bringing Order to the Web. = Manuscript in=20 progress. http://google= .stanford.edu/~backrub/pageranksub.ps=20
[Pinkerton 94] Brian Pinkerton, Finding What People Want: = Experiences=20 with the WebCrawler. The Second International WWW Conference = Chicago, USA,=20 October 17-20, 1994. http://info.webcrawler.= com/bp/WWW94.html=20
[Spertus 97] Ellen Spertus. ParaSite: Mining Structural = Information on=20 the Web. The Sixth International WWW Conference (WWW 97). Santa = Clara,=20 USA, April 7-11, 1997.=20
[TREC 96] Proceedings of the fifth Text REtrieval Conference = (TREC-5).=20 Gaithersburg, Maryland, November 20-22, 1996. Publisher: = Department of=20 Commerce, National Institute of Standards and Technology. Editors: D. = K.=20 Harman and E. M. Voorhees. Full text at: http://trec.nist.gov/=20
[Witten 94] Ian H Witten, Alistair Moffat, and Timothy C. Bell.=20 Managing Gigabytes: Compressing and Indexing Documents and Images. = New=20 York: Van Nostrand Reinhold, 1994.=20
[Weiss 96] Ron Weiss, Bienvenido Velez, Mark A. Sheldon, Chanathip = Manprempre, Peter Szilagyi, Andrzej Duda, and David K. Gifford. = HyPursuit:=20 A Hierarchical Network Search Engine that Exploits Content-Link = Hypertext=20 Clustering. Proceedings of the 7th ACM Conference on Hypertext. = New York,=20 1996.

Vitae

Sergey Brin received his B.S. degree in = mathematics and=20 computer science from the University of Maryland at College Park in = 1993.=20 Currently, he is a Ph.D. candidate in computer science at Stanford = University=20 where he received his M.S. in 1995. He is a recipient of a National = Science=20 Foundation Graduate Fellowship. His research interests include search = engines,=20 information extraction from unstructured sources, and data mining of = large text=20 collections and scientific data.=20

Lawrence Page was born in East Lansing, Michigan, and received = a=20 B.S.E. in Computer Engineering at the University of Michigan Ann Arbor = in 1995.=20 He is currently a Ph.D. candidate in Computer Science at Stanford = University.=20 Some of his research interests include the link structure of the web, = human=20 computer interaction, search engines, scalability of information access=20 interfaces, and personal data mining.=20

8 Appendix A: Advertising and Mixed = Motives

Currently, the=20 predominant business model for commercial search engines is advertising. = The=20 goals of the advertising business model do not always correspond to = providing=20 quality search to users. For example, in our prototype search engine one = of the=20 top results for cellular phone is "The Effect = of Cellular=20 Phone Use Upon Driver Attention", a study which explains in great = detail the=20 distractions and risk associated with conversing on a cell phone while = driving.=20 This search result came up first because of its high importance as = judged by the=20 PageRank algorithm, an approximation of citation importance on the web = [Page, = 98]. It is=20 clear that a search engine which was taking money for showing cellular = phone ads=20 would have difficulty justifying the page that our system returned to = its paying=20 advertisers. For this type of reason and historical experience with = other media=20 [Bagdikian=20 83], we expect that advertising funded search engines will be = inherently=20 biased towards the advertisers and away from the needs of the consumers. =

Since it is very difficult even for experts to evaluate search = engines,=20 search engine bias is particularly insidious. A good example was = OpenText, which=20 was reported to be selling companies the right to be listed at the top = of the=20 search results for particular queries [Marchiori = 97].=20 This type of bias is much more insidious than advertising, because it is = not=20 clear who "deserves" to be there, and who is willing to pay money to be = listed.=20 This business model resulted in an uproar, and OpenText has ceased to be = a=20 viable search engine. But less blatant bias are likely to be tolerated = by the=20 market. For example, a search engine could add a small factor to search = results=20 from "friendly" companies, and subtract a factor from results from = competitors.=20 This type of bias is very difficult to detect but could still have a = significant=20 effect on the market. Furthermore, advertising income often provides an=20 incentive to provide poor quality search results. For example, we = noticed a=20 major search engine would not return a large airline's homepage when the = airline's name was given as a query. It so happened that the airline had = placed=20 an expensive ad, linked to the query that was its name. A better search = engine=20 would not have required this ad, and possibly resulted in the loss of = the=20 revenue from the airline to the search engine. In general, it could be = argued=20 from the consumer point of view that the better the search engine is, = the fewer=20 advertisements will be needed for the consumer to find what they want. = This of=20 course erodes the advertising supported business model of the existing = search=20 engines. However, there will always be money from advertisers who want a = customer to switch products, or have something that is genuinely new. = But we=20 believe the issue of advertising causes enough mixed incentives that it = is=20 crucial to have a competitive search engine that is transparent and in = the=20 academic realm.=20

9 Appendix B: Scalability

9. 1 Scalability of Google

We have designed Google to be = scalable in the=20 near term to a goal of 100 million web pages. We have just received disk = and=20 machines to handle roughly that amount. All of the time consuming parts = of the=20 system are parallelize and roughly linear time. These include things = like the=20 crawlers, indexers, and sorters. We also think that most of the data = structures=20 will deal gracefully with the expansion. However, at 100 million web = pages we=20 will be very close up against all sorts of operating system limits in = the common=20 operating systems (currently we run on both Solaris and Linux). These = include=20 things like addressable memory, number of open file descriptors, network = sockets=20 and bandwidth, and many others. We believe expanding to a lot more than = 100=20 million pages would greatly increase the complexity of our system.=20

9.2 Scalability of Centralized Indexing Architectures

As the=20 capabilities of computers increase, it becomes possible to index a very = large=20 amount of text for a reasonable cost. Of course, other more bandwidth = intensive=20 media such as video is likely to become more pervasive. But, because the = cost of=20 production of text is low compared to media like video, text is likely = to remain=20 very pervasive. Also, it is likely that soon we will have speech = recognition=20 that does a reasonable job converting speech into text, expanding the = amount of=20 text available. All of this provides amazing possibilities for = centralized=20 indexing. Here is an illustrative example. We assume we want to index = everything=20 everyone in the US has written for a year. We assume that there are 250 = million=20 people in the US and they write an average of 10k per day. That works = out to be=20 about 850 terabytes. Also assume that indexing a terabyte can be done = now for a=20 reasonable cost. We also assume that the indexing methods used over the = text are=20 linear, or nearly linear in their complexity. Given all these = assumptions we can=20 compute how long it would take before we could index our 850 terabytes = for a=20 reasonable cost assuming certain growth factors. Moore's Law was defined = in 1965=20 as a doubling every 18 months in processor power. It has held remarkably = true,=20 not just for processors, but for other important system parameters such = as disk=20 as well. If we assume that Moore's law holds for the future, we need = only 10=20 more doublings, or 15 years to reach our goal of indexing everything = everyone in=20 the US has written for a year for a price that a small company could = afford. Of=20 course, hardware experts are somewhat concerned Moore's Law may not = continue to=20 hold for the next 15 years, but there are certainly a lot of interesting = centralized applications even if we only get part of the way to our = hypothetical=20 example.=20

Of course a distributed systems like Gloss [Gravano = 94] or=20 Harvest will often be the = most=20 efficient and elegant technical solution for indexing, but it seems = difficult to=20 convince the world to use these systems because of the high = administration costs=20 of setting up large numbers of installations. Of course, it is quite = likely that=20 reducing the administration cost drastically is possible. If that = happens, and=20 everyone starts running a distributed indexing system, searching would = certainly=20 improve drastically.=20

Because humans can only type or speak a finite amount, and as = computers=20 continue improving, text indexing will scale even better than it does = now. Of=20 course there could be an infinite amount of machine generated content, = but just=20 indexing huge amounts of human generated content seems tremendously = useful. So=20 we are optimistic that our centralized web search engine architecture = will=20 improve in its ability to cover the pertinent text information over time = and=20 that there is a bright future for search.

------=_NextPart_000_0000_01C85294.643E2190 Content-Type: image/gif Content-Transfer-Encoding: base64 Content-Location: http://infolab.stanford.edu/~backrub/over.gif R0lGODdhVwKaAvMAAAAAABERESIiIjMzM0RERFVVVWZmZnd3d4iIiJmZmaqqqru7u////wAAAAAA AAAAACwAAAAAVwKaAgAE/pDJSau9OOvNu/9gKI5kaZ5oqq5s675wLM90bd94ru987//AoHBILBqP yKRyyWw6n9CodEqtWq/YrHbL7Xq/4LB4TC6bz+i0es1uu9/wuHxeLhjshvu9QO/7/4BUeAcGhIaF iIeKiHl7gY+QkZIqeQeWlwgHmZsInZ6fmqGZl5aJeZOoqaqPiZianwgJsbOytbSznqKkpYR5fKvA wcJglaK0CcjICgnLzczPzsvJ07e5o7x5BMPb3N1KxZzTzQrk5ebn6NHT1LChl4za3vLz9DOFlp3K zAoL/PwLAAMKHBjQX79+6aBRs7WJ1Kl6KQYUmEjgF8SLqO7lg2aQoMeP/gMNplOXrFonh70wXhhk apHLlo0MxFNJE00BfLH2HVzAgKfPCTwl/BzasyjIgyLPPSspyxq2i5Va7dLVsOpUlKYs1ty6Jc9G cgKNAh0rVCzRowCTolPIFJc7Rt2iGnLFCZZdu1TdYTvkiKtfKfeU/SsK4yxIkQjNLUXm9hohYPfw vcp3bJ3ly+tMOnVc6s7fz982hnVxtidatUqdtW3IV5JcY7XWqZutULVlzSc589UKuncPS4IB1ijt 8zDSxOUWr357gPcccK9ij0N3HLE/sCPJsWXMMPdURDN9i6dx8wAz4TKCmjV6GmlC291ZF3Luxg6p jYLNVd9vvXr22gvh/uLdO73QN96BKbzCTw/qGVacQNalxpaAjnn2hkY5BfcPQes5aNpHEY6kHHeg VNgcgiiioMl5ORDH3lFqIbfdaie9Y0AbpEjXEYcuOmhciBJKE6BTBD6U4pEfmLcgg2MR96N7ipFE 4Vs3ohGZOBv+ZEKPwhHUn4hCLmdilUjSQEBFaK5SCIstdvjik6hFOWJ8DpE5RnlfZSmWCA122B6Q coapGVZ2lilCTIrw8l2Ruxm4hibkDMHlgx4BSpuYnZERnU5azjDphwX5txZ8U46CiKEdsHSVK5K1 uupeMTk6hgEIRKpDnz6iJSpygd7WVI2ZgoHTMqPFgKtZf4qaHEkk/hKJiKwpSkXXXZR19+tmejHq ixmQqgfEpMleB6ZypQbLRWCcMukkiPzxCl+z2RqJ5Gs4fWJLbJjhixu22m4bxgG2tgkuu1+6q92l JF5b57n5DHarn5QSrGyvbTUVr4UH3qRRdBnWll1qB8+YcInxptQFwN7+MDCcEy87Z7kmX4EPiyn7 8CmMLQNIZ4XiaVxvx9Gck9jQH4fM7L0kj2nAAFkouUOPoEr8ZdEiVzwgeDKbx+bTTbqp69Sjvqyw qYRAS095OGGJ3U66htTuuGE2OyA2hUZBqwIqe92lcWm9/d/R1TBXNxQrOiyp12966TfFI+di41bg AF0O222HC2WQ/r4mXSfGUMRSc2GIfx1qiAZjro/V/PorBaSGP1zWyopbGnbcO9OtEoYahuXi66En 7nacgI+NFeEIfH4D1BG73TfY71U9qKmDL1FAJ0smgXzUv/OHuaDCxywPnjmtHZCbZfGOfLjYuUw7 Y6l7rwTKw0GMPc4FF03b+kg76/4ShdRqPNfm8x39Wja7zOlPXtu4SSaCM77/keBYuXqbu8Q2t0Pw jxkvAIAGNUiY0FUuVBtcXPOONqQKRu8ItEoAeqynN9Exr4Cou9oJU5G2dDmwBOtimbhM1zhgtQYJ 5lkhChawwSJq8EHzU54EFbBB+xmtaj0kUmfC876t8QCCEOOb/gjhNqPn2U4YtApf647XwkoRMHgk M4TZduA/FhRxAhv0oOiWR8Qjiqt0TyQh0ixWoRkOwXM3/FYLkxg77cEwQN1D4CT6V6vBBHIF5xvg 5A6JqSIVwQBW3BID4viTEFZuQ2DrRxPfk8c5Za57vFjjD7p1ONjp8IyMy0ztfogK4OyjgzYrYyEt RUH57O83xXskB3hSxKGE8JgAAIgRA+LJEDJRg/xwZjmKqJ0NJgCaOkMkzMwFxDYKAYs5lOQWt2dA E1IxEP07jzBbEMlXfiybs4RLEJQkxA30KIRJNGIy9QmAaNpxlACFJjn4yQx9YuYWteujTJJQiIBd UY5tqx/R/qiGMKspzEZ+lIOCxvhNXWaPl6RK5Il+EIvqPbAonERWHYvZTGg2caWiFOgon9lPmtK0 iVCkkRQLlASnGeF6nxydIUcIT7n5UpU4Kuk60yPHJIaSizHsI1JhQM9OgVOlPfHkR7QqkmI+c6Ay ralNbfoMl2IzX1FMIzamigNMmvRhrmQXHZ9K1HdZVK2/nIPW6kmEuH4Uj3aF2al6UFJHdvCqRNGq Rzy5IX36c59mFWtACWpN/I3sWt6RKhI819FIEpJtEt1hAfXoxcf5YU0qXKqxICrXcZaSe1PC6G8Y qLu9JW8gXmXmEZ15EGTWtIn9DK4G91FZfg7XmqfcV2Y3/sdWG7DyA/ocQTuP4tglOvG1ycBfaZ/S B+DgTQlA1aLs1FdCx5WCB6zTXVMXe0SBABeavf0nfG9K32lis7IKQe5B79oOsqmOCPCDLj/hyEEP IPZDoDLuLl1bV7RS6GojlcOa3sqkAN7WnekbLbnoJFsdpFB8H+xb36gZ036uFDuTtW9k+1nW4S6j uMcdrjbjo7ki3eGcQLgbXzEQRwm8cZMFpsCBL0wQrZKYdPbbx/32uy9C0cGnLAyneHelYVmmcbDO vUTuQgwh3fKzxAYxrjkE+tWD4fSmRbymjJOL2f46WQZGTFJqBQwACvS4xxcIL5EZC9PQOhFAdtVm +zjn/gZ6dtaj4hxvLAXtw//KYE0KzM8kl2hII46ZzGAVLotfvGkXtxgAi0kzMvSL0AfzC1aEdoFx 8WwBQ3vgxxXQZyfxqczdJtPLtxbxSk0cImpmGs3NOPOo9btBBAzXi5qFA5SRcLOIMriUVtZfc0XA SgXpJLSUrl+GrzsdQBc1X6WmMWuwMm0QrNq4EkjhjjMQZzu3G9275mC821vk3IZZ1mj29VjVbMSC GrHUbqZlGzCpwkNLOdFUbrB2UdkLHK9AxxKoIXUmmeGJclvh3g40uPeoXKvsghF5EMAMzk1yeZs7 zj6hNZDrDNN51/rWL0VMdVX861UH+6zItaZBTQ1h/kerQd0dtXAWcTZXWC7ZynxcWAwCzIBIh0/J d6R41C/+N+wyy7JIBzi1skW3cp+85GCv86thndKUwlSxaM+1dSwNZrBE9tKavjmLh81vjru5j0zD 0Zwf6tcFI5mi346iLgTOgmUzAHfZ9RjVlZzxxp/OlBsPN8/vvqhYOdyNYSf5DXH1xpZ7/uz0znfM tedrmwtb52tetbE12PHBdybVZ9DEutXF2in7mYu0Ky2WC4/BCjBSOoqnOkWt3nisL+SyHV+ujRV5 gsw7v1MawDNMV45Sk4M+1/PmrT9G70zTY/q9nTZu3ZEf8M54HQuGp8GQP6u4ov99VNi9zHZ5ygKI /luAkfhZzJL3X/ztGD/yHJc/1KJ8ldcIlxd9zpeARdRI6kUcKtdSuZZ2kJVM2zdfpPd9YSVWSiZs aKYP+jUNxSZurscXsFcGbhUEetYe7jdU8FdRjGZeeVUCnLUSPyMO/cd/OJhN/8dmkjeAefFxu4GA CjiE0cUAWgMW11NMKwdz9NZyEAghL6V9JfZbM5Vi+MVpVzh+qAdqWrdT5scGQaRafNI1B2d7z0ZK R5dc5Ud4JqAJsEYBG1Mt+SIkdJh4j4dWeDhjkvcrbYYXVqEbptAIfECEhFhyFLAinMIje6NgL9de YtZn8eVY6CBm1cRPZCUkHzhs+rR6ALB1Hkc3/iU4Bs/FAaxGfUv4hisHVLIWVLcHWI6nhzuVUUki emKXbjXYMZaRg/x3h3noK8hnd544ggdQiArIYyR3AYWTJV3iJquIa+7FWydWHfYmjZY2iZVFU7Uh bCBoTbHAjSI4bnxxgP+SSewWZD5WYOdWfSyHYOyFb9lmcVWXhgBYY1OUAm5liHHoCQHYi/wof3cV gMmHLVVBjAkoAqumAbakH8oYVLqGbVMnWsOni/qXh3s4eSPICOdnBYA0drV4jmLHap2nckPRJUbG VUKlbRCJe84jgAckixzgOc5nDNVSGf1Yk7L0ixU5gKBAkIa4AkW4AWG0ZVzWZQ35ju/HbYyX/oO9 mJOaozRr0FB480ileGdHZBT4pFggAY2jp2ho6G286I+mVjI+NwIoQ4R1AQsAmZYVCZCU0ZY62ZSi wJMdCWfm2AGbopBGaV0pKXzktYuQx2Rs6Yc/iGoZSQVuNXuxVpdUuY4emVVVOZK4hZW6dmTgN18z lWnhJ1x/yZIm5JIacIIEqY9ryR2kqS9MgZNqyYc6ORmhEJOC+AQJeW3tUpQOuTjweF1J2X9fCZhd uFyAWDaPwoBSqZjy9oBPSHRs11XVZXO/5m+Wxm+ZiJOn1gsLZYxySXLTQi2qmZpM2YdvmRu+qRec uEGoFiuF2VaIKGkKWXFSd5t8mZTEd3U1/rmW3ymMJOiZJ6NOuGSdiZlMp7hMjol9aOGOjwVmzDlW 2GhTBjWPbSY410mE0lIVd7Gd9MmdbvmdAmmfWHGfE2E3dwl1eFmbD8mXURKfOsiPTHkXg2lj5zkF 6UUU5aOOYgGBzfh5lFaNJ2ZfCIqN+IVfajZ+alguSnOMHDBhtBhdP6Equ/CHGNqkb8maf5gXulF5 iyCIN9aiOJA2Txc0JBqPuSmRdWiTqDl5cDkm1HljYIgMy+hAPZZyVWl2JSmgZrQ8XuV9YqWgZGan WNebn0h/I0Ar9ESkFcATg7CkVOGkiOqkUCqlGqotZ6ox+CkEiKcP09Gl77ksJiqPKBqY/irap6AY qVzwog04o/ImHMfpZS93ksehfZPlWGfGo562akD6j8H4cWPpAbHAOl82SAsgF5FhqFEqoXUBpeEZ rIz6KrBSpeZJAH4KBbiDi4B2MNLabZgKpl65mQ5Gq+ImmJ6qrNUZnIl4T48ZoKn6jEx4a6BEjbzW Z5dJc8nRb3JHEtHJX/QYhCPgOXulhEDVE/YhLffBqMNqDeAZsABbsMjqqDHhC+XxLFMQOTa4f9MK n9SaqbqJrf44phfKrVxXpVjqoke4kFtFb3FKp/Alhe1ZelWYgZfonGflnJrIbz1okeCIkeKIAbK3 AIiIhEMZEHjgq3OxKtdgrELLKhuL/qww4a2+cGN0U7NI4LDQ+rA3WLFhKqbHZ6FkmqF9ChOCyLRh EEYg5hGmsWqoaq5TmJyraokqZmmjtIHY5AziN698SoCmFQLVVliUw4629RGI4q90UbBDO7SucrCR obVbOyZcewSSIYdQa62M6392OJ/aGpBwCYQg17ENm4xrk66RuZWcW7IKNhKSqBjU5JzqoF9hImow y6n1GojTFmAJOWm6RpT8IREJe7QHe7u4uyjJSrgJawcFwKz9YgVoMxkI1YtTO7VUm61jKoDeWab2 GYgKGwc5WzDtB1p5SVcX96VRu5SqO51fBAJQFpSyuSOyWz960AiJ8rO5u76C+xJI/rsHBACOX2gF +ai4xvu4x/u4yXua3MmZPmislHumIgcH4lupIWqbUmep1Rq1yEuRavmk8nufhzsBM5hueTJxeuku vlu77nu0HtzBIAxytWulFCEAG1u5WNAKZ5mWpbm//Fuh3YmofrsqvGsHeadsdyk0ImqU7Ll4fSm1 mxqznVqAzXWzh6g1kna9b1MRHGy7P/urVzG4BLK7L9G7N7a6vTDAV/CrK1wZT+vCm1qaMGwtq5mh rwK93woH4EOpIKrDPWypEsvAjtuP9AnBRdso1NZ7E7DGHKHA6cDEPTvCITzI0HumI6wHFBG/3fp6 oQgY6nuWM9m/YzzJsrCAV+uH/sXaqDWcxnEQlFtaNBbnniSqvdZqkymqsb8ZExMsAd4Eh+kZfH5s DrQbyHtLyIQcFYe8B7+rsZR7q4b5yBM6xmL8wpJ8LwtIxhg6w6l8n9lwWoCaf7MxrUm2wI1rGw2M hxlSx2W8ooTiyxlgf3CIf7mYcRELphIxERucy+q8zj27wehcAAOAtVQavV2gpEyqnaLpllarmhnb CZbcqZk8blPawVa8yrHHMYsby176wxK5mxf7wMEYwaBYxHpcAVrKxlw6zS0IaBWRzuv80UmbtBQR z9/4vOH4BREqk2hJySwdgP+czPI80IrCzJxMB3iC0LIhx9U8kY+3g8QMw3a8/sjK2ro0UwHgY791 uNP5axln8s4TgchQ7c7ojCYEEAD9nMlR/KhhoKT/KqyJ+tU6+c+L+re368G9WwABAAkYkn8YrdNu nYY+zZuB2bxRGsCqTJb6eX+3aINtrdT4K6bZbLVDHMGseyf00rdgLSD8/NUvPbDKnLs1bLldm7jF i1ZLzdOYvb+BDdSXTNZZLcIGHXE2NBZcLJr8C8bcO8wxHNTHuqGWZ4Loy7dkLbAB69i0nRcLOMOB S8NVXLu+G9psEDn2O85+jTDXDLnL29nhadfNSrf62Se2CBuVPYd/7dCAHbnbOtgnTIKSnQT9qr6J K6WJzdjkOdviabQIi8hq/lK/lf3FqI3NP13MMgy48yyIWkxtwrmmPUEAXEy8+4gvyG2aLK2odT3P j9rd0lPLr2LeDM6ouS3UvP3Bh0wRwQAd9jLd7+3Cmy3JV63d3aq1v9uGtHUW9oEJK30vLUy1gZ3c zFvGyx3hscIG/B3b4I3Yj93gm1BErS3TzI3G9AxG/X3hkbyHw1zk8g2Miw3TBmvgIhzibShGlCMQ 332odkGhR651SV6f8rzb/WKeN7wGT52wLoG7QUu0Ze4q16DjZY2w7yvSTs4Nc+HV/+3eGh7fA67l L77MGFmYXjs5IFHi4X3hVz7oWN6kS/7ZsQKqX6AqUnywOO63Os7jvL27/iM8EffNDcXQ1Ss95EjT iYQ+14mt23ds1j9uAuk0DhIk28Srk4vd6vn81WPd2mdM02E+B2Husy0x0+yLu2r+xLaszr77u5fu Da/h1ZyOUN7Y0pxN1xF96PVd6icwvZm7Hz4DrKsezPr86lmeqI8962j8u968BjOO62O+67nb61P8 67ncobdD4+Id1sWm2Nr+6uNdIgF93p/9vr9df1/RxukQyDUutAM737bd7Xfs2oaMyAXy5XMgAB4t 5hL+6y+h4xEf2+uOyPDMFYdt7Hfxb59uEtuO5/Sd1eV5B8N+Ar/Xx+3pD4cd6MRa7xB87xKd7xgp E2zYB0xM4x88043u/uszXURSXPHr/uZ+0bMBL6WbCPOGHtODKekEnejAzQE1hNHZEebpa+2PnvXs S+pnDdqT8NS0LMgVH8IUT50J79tRTfS94e5d3ZpJr/QE/+izTul3DQOTGq3/7rNXofV8T7Q0r+/A i5GrAMhNTO4Rn3m4POEYH/X1APB8y083HvkNzuXozbuJzu6PdtF/TRtNfPXIeub4bu4Ib/kk/Lsc yvhw0NFRDdIjjPgknPZncvIooveXoHpU/vLX3l8FL/drzvVqDwNog9TFpwC068QLbuP0DfrtS8WX D7yfKg8dHfasbwCuv/oUjioWcOuIcG7/6vJ9L9B+3/tPf9aoHwLC/k2T4/x42m/4NS76unvLij8R A7B8eoARAxD96Gz9UA12Tn0mDI/9EMDklMVcA/Tm4PwPOURyNEv0VNM0BF/4M8C5PmwMKwqq939g kPJCFBGJY0K5XCqaCafCksPdrLJYVhubYatf6mVHGHBvGIJQvWa33W94XD0gEAr2XX7X4WsEcsBA wUHCQsCCvo6uq61Gx0eayC8rnDADi7E/w7gLF6MjUCRRJlKlAcyws8lVylbWGssc1DyCl0oxg7TN Xd5e357Ejt9h4mJjimCORUgRF2eT52bIrMVXKsyCgGOhWpBP0aTR0vEEu9TXViwvV3awMGwVGFix 7Xr74uSN+33+/n6JfGXo1DGaNrBaO4Tzrl26dMefjymePoWiCM4iRQEEGMaiosojho8c3zEc082M QkwPVa5kk4/lS5iEAALIkZDRsoJaDqK7dW1MzApXUBi5WDSUknBHTk0RGXKeU5Ek77Sw5UoHUKwr XWbl2tUHQFQcbfIca1VhrFk/u2q04W3ExG9E5RZBQpconh0k9eLiq3eW1EzQTpplqMvr4WNbES8G mkyCRqYgyVYxWHZVJ7QNGTPo5JbExLoVRRu9WEePX9RpSwZgIViaJEpoNs/2FYz2bZWJfCzdGPWp ZMzBf/vWIbUObsxCVxR5C9f5c+jRP7emuuWymAIDcG+XqZv7/vd6fYTkjRzL6XngZzliIwM+uefl 0uXPn9h8xf3Xg3vuAN8/jm3/AuxFvDaWIm+v3ogrDpe/9DhOwAqSwwk+6vCzr0IMXXvkuuIcgvBD IAAEcUQ5NGCAA0AM/Ksv35hS7Q7DSGQgorY8sdCIC6PLMcPWnqGGMHpk9K8RAoU0Ug0T/wGAF9Pw 0uPJJ+2oYwDtjgSiphqzkIbHG+/zccOxOszOyn4qM/OmSMzs48yDsCATvCQnWPLNzWhMJ4Ytu+Sy hC+tg009TGKkc5OcCs1pTUMhGTSrOBe9TYAEJSQo0Zx2Smi9hgR1FI420WQzTU9D7TQDPkYF1dRP P9iUn0ZX/r3NTsoolbWqddzRQVNXg5h110Q74HXXXIlpNdjZ8rLEslOTnSw29RgsQBNifxA12VSn rRbVA3y9llpruzUgWkOGBZexASIla9taLVvPw3En+PXdWbWFF952/5uzXtx4K68jyZC9RRVM08I1 WnS95RZbhKndoOCEGU4W35buhRg3AYzFUtJlrwNYloaqhHhekOPVIGSSD5g4RIlP5s60FRV02UXj Bh63YYMdtlnUDWjW+WCeB1JZzpR/7o9lc10eCZs7oP25ZKYLXbjpkIUWV2gB6XDywAahhHEAbaj2 oeeaw965YA9uHlvsNKUO2mu2hYb67UfKhpteladu+26I/jtD22ywz57BA7/7FvzMrk+2G2/EwZ0h v7kbd3xWMpTG9/DEK891cVD21jxwvjvfnIatuTZ8bctLv/yDUZx4fHXWqck08tFNlz3aC+xSIgoF FOD8c893912eM0qisvDJSZ/9eDILCCEJ3HN3XoEFWpf+V+Zeg6UkASRvl3Lk5eBd7O7jsOCbJ543 H3roex/8+9+xAAVHPmXQIXTitzc+fDambwT/IJR/63YomG8B6BtgARdgwAMmUH9QIwdSwAE/W8yv DgKo37i4xz93qU+DnsJghGwHhQCeL3cFJGAJSXhCEyqAfesbVRRA6EInjGM0nxBMbIQnOohdEH8L 5BX//vxHlxeKEIVDPCD0EnhEJCbxiEZEIBNTOMQnOm+AIjzfC60YQyaIBoKSeB0FJ6bD7q1wgzrD 3/j+Vz4qRhGKa1RjG6E4wjTGsXlXBGEDHfi+6tVQFlvzYvE6KAQe/qqMywNg89RYRCUmUpGLbGIC 3dhGOMoxd3RsIEW2KD/sTLCC0QJjGEXQPjF6iwDak934jgJDIT6SjatU5SMjKclJwpCSpZjhJW3I x/u5qpPIM0ARVBjIglxilPjrBhBvl8pVIpKRjHSiI53ZShS+Uo50dCEptJhHGkhwlJsM1i55eYBC 5g6UoISGGAjATdP18ozhpCIr3QlNV6IPllV0HjXt/njH0FRnj1MiJbG8ebxeNkGKBFRg47D4wHIm jX+EvOJAk7lMiCaymYiEJwmlSc9YWrGSD8QmJruYw1x2cAbsrGg8p0jPOmaxLuXMxTn5t7jUBfF8 76RpSd950WnKUpYyvEh9rBc8Pvrxj0LoJfMOCdFmRjGNKV3C+6CRiYWuM4QCdKMyI6rEiTbSpgOF pT0Pik88xg9gYxhAP3P1T4CCU6YO3epJl7rTLG7xAnVAZ+m6YMy1zrSte32jPOf5vBDa856WzKP1 tFlWoQ4VCAUA4vNOeFVnUrSv9YQrQkGH2B2eEZVwfChkIZvVveIUsJTVKDkIK9Z9DrNeaJ2d8pjg /li+llC0lb0jn+bqUvzdlXl5ZWtNfRvb2VLRq0xV6Wk7ekt+JlaxEGnsUZeZ1Mmi1JorlcEOcBg+ hpKUs0dlomcj20jJljS45hvudE+rRwli1n7LXewIePvb6IqwtNaUa0tDmjjdvnaqeoVvbPvq1r/O Uad2vKY+tZk95bJ3Aow16na9q9X4ztKyM7BuXS2HuUJO1Z0Pvipo+3tSAJOXtNU0LUdRm16zroq1 x1PrMR0MXL+KmKmWZGl7XsrQzVL1w/41aYCH+9Wm9pSGhs0k7Nar4B8098UdfmZ8R0tL6t7AuhjE MBph+9DuPtjD4f3teJ88YPOaGL2ZvC64Vjw7/vcacsnijeN848oc0Nl4kFJVs455vGMQB/jLgi3x eU9cZAQfGck9ULIJvQvdFMqYxBM+wJRv3Ni1bpjDSP1uaGOMUTD32acRxMbwBD3oIRQaz4m+9IiL u9JmzNVj4fthOHi75lHHWralbvOAF33qsB43tYE2832H6l4NG5rSXJ61fIHsVKCuuntV1q4qrcrh LY/ay6amrVFsOb/slNmfvv5jUV3c263mlLhNxVGclY28Vus3jrJmN1f1bGs3hzms8fAo9iy8qDOj ucHC/myTSU3t4qa6ISm2K52D3dlJL9LD4aY1ZTN6a/NuOpud1nY3uS1SUd/Z3YrGtW3vcO7j/qW7 fPvVeLuj2XDhwhvIYM0nNq/XxXvTKd/6/jYRh83Kt0IZztgh+IUNzl/uJhy8EGZziFPO5yDT+MRj yHbM3zRz2Xmb5LIW98rz+QKyYlDkzb5yyRlu9K6qnMBCNjCgnU4mqJsO2F3Xsr9PDnBy71yYPa+c KdUNbpsLXZEL7+94A1vtgA+Z08L0dK9B/QNvbwDW0NxAno09Y4QioAYK5V8xXZ3jxZvc0mA/Orx5 alzbBi9TZT27ldJeOmArPu/MdKTqiy3dgMuP8mWUerMlneVDV9rt8Jz2wyXM8sJymn6GP7wPitoB vDNeA1XXebk/ANXKE3K36/b6jnv/eM/r/pzs9BY9Wfu47eJ/7QgdeKxnjej6t/v+oDQMnmpf+kFD SlrvEtX912ut/rHDpZzdJwPIVXzxbiuC5Vs+ElI8DkChA4SexuOA8+GAOtqABNAAOOMA9cosvMK8 6vO668M/wLM2wbOh0aO7pwNAkcqWASRABUyEBEoEBQgG52HBCOwAJOgDEcSvC5y6qkIgaNM9Yqup 6yuvjgu+iQsq4gu/BUMADQig5UOfBHQ91VtAxWvBJZRCAHACCIxBDkBCDbiADahBxJE+rks+zfNB lGO+TJM3MQs9ZzEyTiJBKtPCF6zCAmq8A6JDOlQmJyTAKLTCJLxCARwZcyo9vPG26RND/oLSwfmz KkTrsjL8u3gLPLkCqo/zwiM5vdJBhCTMnTyUQypMQQAYosbTxCWMQiiAQAgEhT6ouG+6wTobQw0s Q5TKPnlzjqrADuuqQIszQgoglT6YQw1YwV9cADp0ok0UxSqMQwDAQqQwAj4QxLshRNvLQdwzPx6U NlgEOIjzs1p8HVw8Kzf0IRccoGIUxl80oT2MQgiMAlNMQvbjQg1wxrbBsEJspwzsu2vsPKTDtTTc P1yYElX8P10EmmR8nicMRnL8xDoMRijkxGLsRCx8n4XhRXhkG2iMNGlMRCTiu6K7vzNsvg+8HiIE v4BUkmQMoVAcRyokv1DsRGOUQWVE/kU++MfZYbaGYrt63DwfE7vPaznnu6VbnMgRscTSSUcl1EOD vMOEDMZOFMcpbEkX8sM/3MIHoTLUmUdDHLodVMTdI0POE7DSOrZ95CLhoUQhEUrU27fMayXmizi5 czSqhL9Xu0l77MpaA0I0nDc15EaZxLdvxCC1Yrvc68FSo6R5Cx7/S6tTOrigw8hqxMmc68ik40l+ 5EaylBGztJy1CyDNm635KkzsOMyZ/DkcdMV220CxszrQ0yNJbLoiHMm/1Ex+Yz3B7Mpxa8dcAM2o q8poxDKhizbHTDn82yi83MaxbM2AXDs7s766BEvbnL2XqkgMTEvSVKp7xL58jExt/hTL4WvDkXSX mjvEfpvNWKwjz7zNP1IeorBI3mRMrRRPnGtEWQzCSAS0yiSRy6yczJRO3mu4zrw6oKpPivyfyxtN ufzNsLNL+dQ1oPJHoASR+0yc12zB1VO4reTPbCw3VTtPHFPPxZw031RO2oS7Prs6IoM549TF1JtC qltOjxQ456Q9uKwz/ZzOk6PLx7xO4CNR4txOkexO90K/aXyuCg0xwvTPDO0gkWvFDmXP+ptLFu3A 1Ey1Bd2mEzXCEVg+2CxNC4W85nQ/H6rI3SxQEM1JBMVOifMCymzQD3lQxFE8xWQy8bTOpMPQj9NQ zeLQNXq2tmtPawzR4BzRD6w3/hj5vlzsziuMPxi7UbZ0US+F0cSUUTHdTxvFR8jM0UBFrsjpSwfV 1O7ZwiMATGqM0z2LIacSuEZltQ0FOpsLUoxcREl90kckO5ebUkL1xoDUB9QJos3c0vUzUmFSU6oB 0/dyKBak0TbawwONTzO9VJBMripFMn3gjHAAVSEVVY6b0517UQucvjctoWTY06FTpZNUUY78vVnU tQOzIE41nUYZKTxlOEWFRPmhq6GCKW6lR1/8RCYMRWOdLWTluEr1wGub0m7UpXWtnGidgKLCnfIL z/eEu1KtrlPFrjtVVW8lQGBEyER0VTZyPb9TVrCSuGadWIAEtYRV2Ah9RbDr/s/qAZxfrdfPuDvY 4ix+9cQTYkCmrMIELMCazVeGPEEGRMYqBKF1vEJllNU/o0x1NVnSYbDvREQKtdY9m9MrpYnsAFC2 SdVu5VmE9ESN7cWD5AOfxdmULMhP5AOvDVoXVAKjDVl0BbSlVbCT7QHXEqgZhSQWXVQT1ADcKsGK JTkHSwYCGledJdx9lUPDZUl0bMrFHdpgWIINKDAFLbKCLdnlmlu69aUsZdWofdjgjFgOAFa3qdjk vFiwTcoj+kWkXEhEYl1zLMeGNEaH5MNkbFu37clmrdxNYdNPa69w2tWVXbSI3VvMXbbsGlaaLcec PVy0RT/XXd4+OFzZnV5j/lRGdezDJLzdt1XaZ+3UCyKAjIPavRtSfKRaECiS8KmyHOtYpUTdsFVJ g+xE1UPdtEVc2J1CZDVALKXdo4XE7fW+uO2g4gWCFttaNotXqk0170jf44XUm1Vem11K0x1c1YVg IFXcUcTfKVRGoqTd7AW9/w3JHsWfAQ6C5mpYOPXcyhpeTETf0LzA0uXa7rpDO6xghTzK+7VffTXb 5Z2ixpXdF3LJc01ashJdAeFdgyXBlAXe8q2tGnOMtPo5qkpefc3XsdXf6M1Z8itbFGzcDEZb6w3i 7F3WazswrN1U/kGr8OXcjCRfOXViENgDKM5NSENUKBpG983Yr7XhrtVj/tOt4ggOW4LM3/0l2jFe VHSlOCMOECTmy74Ep31bUQR2YlMdAIAopQZG3tgCUkacVOsEM6MFYSIe1DNGY7whnRJuA6NC4Yji 2MEUXlSTMh6YiYM9HRi2WKy8KqQcX/fEWzOMQeYMS0Elg0Vm5FoeFMyFuncd0zeOPPkRyEtOnK3T ZFnj5I00Ta9ywCEOYU3qXbVJ2FSWA2qVTRX2PfP9FmSYCZ9jxavMqoi6ww+F13LlX+E80+5jzQCm GvEIZ0G4WzZay9gLgmiuu0xWUn/OQNMcVXPV3vmMmVotVCsxEDsABBExhsBU4flqgzmuHJrEV2Fj Y71z5ZuSpDItsDHr/pi9dJRGNgZY8RfCEJSKvgf7w74SWWDLmWaDjlSu3DiS1knt078StTdvZgxJ 0RI9aYQqUIyX0Eql4oWYRhy7szJnE1+H1UhGzKk//Tx7ZpD+K2b/WGlCKGoboY9N0yfifer/Q5ET OeZVsTx1w8GDdsWExsZYFeXJZA/SG+qYWIbpqIh7+mukpWW2hhNwHmxHQc/LQ6aPZtI27uV/7unr NGkiy4PCy+eXeI+YdbUMO8265ijB5oFtSxmwHhG3lupXMtZO5ulfxiLU3MegHlSv/mrDDoTOyOww HOkQTSlalgSZQbsfGG0QgU4DLj+QJmer3k+sDtja+kj+o5IE4wcL/iCCxLZJe7TOmVDNILHstilo uNZpvPXk8QxYu5bSNdRdg32I/BIoqmaABWBv925v+GbvA5Jv+nu9fFhu1Go07faa2hvWVWXsJSLf cv7keJNs4ZvE596GF0gd8Dyi937w+I5w1iO1ZNDbn7qAKj0FKQHtdinoKfZu3FZtWGVtQIWg12bD /SaGRXit7nJvQYhwCJ/vJKJgtCWtiMPuVaERQDSYnujtERFu6l7shIvn6h7xcfPfvKQF87Zclsax Fo/vOGhvCYBx+J7w9z3Gqb3dbbQSjHmBkXGLPtGP+TGSjqZmECc18AbYhUbaWbXFbCvloJxtaTlh KC8GKn9vGX/f/vQzNWxVw+CekKGwC3YUWKDmPi/gcAHpbyX9bwBn6iK/0Woz8CF0Vr3eheVxAgcn BimHcDxPop59vTMcTkpA9O3Ikr4uCsgtyZ2MUjEnde4oc72K65s8bTKFq9YezhNPaSbnhS44pvnu hzuvcoN04ye7ccNCDr5Oz7eusw1WVkmvFfeI0Xb+rqwMV+Vc7SPv8zbn3tWSc28zIn4I9jyn0H/L 8p5CL1fnCr1pjgF9N0jnUh29cGFCDtGM9TP/L3fn7FWnRe38SZDqhRCAgjpfiWCnb/rGqj0vUmb1 iugW0Kct7hkndj7X8gunDZgy7SVlbJGOJ2wPZlGXPRFO8UDI/lXo8YdNL/gqJ/c9N2d5Je90Vwnd Ct+NxXc173PsvgSidngzR2219Ks057Nbj/fB8/d/N4Q0+3WYQPlOR3jqnPjtyyaYaGDw5HSlp/Jx Z3qLUujtI2909gp7DVNG99DGvHYjtzr83vbiLHpCGIER2vR9OHmqX/r6XvlrbXk0VYmGP+Enf/Ap j3uUt3JYtXnDevnLlqpolHUmdvcfK3FmXc3K7nZCYLCS94qqP/iIb/qsFnUMh24nl/Fh+Pu5f2Uu LWN1BzawP0SIF3vHxvx3z8atFz5Kh3xBWJzJZwm4t/pWdrK6F/y7v4de18yB34Tbj3GV37gC11Eb wgp5PPx7/v9uEW/9J9h3IZS9lnr8SmeDgEJ6rgB9wKd7ln/9M6iHSwd3tz8G7r/8hIfl4YWFmEBs q8w8Pb3o4ybDjmf8hsYeOLfMRxZ4lRh+ce9cCFBL0qkuzirtzhOYIKKImAdyHMZqMC8cyzNdswd4 LUvd+z+DJwwGd8UdMjlZUjSZzyY0Mp1SqxYwqwUWDCZSqKOxVMrMszmNXqd1zvcbKveE6tMSNWVV sQyGQgEBwcBAwNYhYqLiIgMAwOKIEePk1pDl0ZGSplqTk1wdyRQVCp8LpQ8Kgg7P6emlpZFmUtsT HYhoFd9BQetpgQqYWAYbkoTsMXKyMTEnGpzG3KddKJ6e/kopoGBhL3d3q6MiioT35CuRULJSc+ez NOhdHumVHzkDy8hqvZY5rDJbnC1q8a75MaVPS5cvI6CMwcDs4TqIzNxQfAbNg60ooASOsjbvT6BB 2w6SLAkDHCJxrExmWcnvXCZ/ES1Ge1dNHh8C3oAlkMQSiEuY/WStu/hBSqhRBA3+lPErVQ4Ob4jt WLYsnbpiWqtK5GQRQ7SA03B11OUnGwEBhpqy7YVyC4sLbWe8RIeVa1F2cMJuhHdzKdNTwIzNlVE3 JtF/NMUmzfWx8IuEVOoIc2i5K+aZzvR+rcVXo02yjs2C1KYWMupEb7PgUJC6x2G76axO3IvR3a2b Vljw/molzvVr2DDqIssL9jbSeLsDN336heEwNVzvUk+MNzOTzreRj01Kak8LtKeDk++x+ocBDisL x65OOy8Zo9xDN/ZYytfv8sOFCp2tOL5noOWmHGmFSRaMVHBgpxmD/1mm3QV8RbURR7nsFl5IhDyi H4cnbZhFJOXFhok/1xnnyXZHdUfgUpN40dN6HY4ok2IAqZhcWVi0RYAKz1UW3RrTLVGddUMuuNlX n1FYYY59lJbWWh1yeF4PKrXVHpEm1tiZhDeumKNZjHihSoyQYXmXV4vVNKBuBbblBVRhJDhVg3Vm VhGExylJITx52PcRWgMIIKWUVNZwwDhm7sdPLFky/qjdmnxW4xhvkMBI6AtBMYrVifJ9SSlzJfW4 0I/NCJmle1rZyZltXSbnHXhOiodph4bK4AVhJGnKn2ycqlobgABFmKKkfQ7kZkpkkoclYpv8WkaS xLLJopNsjYmgRUeuemcnwarJnYCvEnjhWRkOSqt+tsaAg08mzdior1qmGa2rfdX3nVm9bYFCu8su yiu8/s2LIm71ndDiT6NGpaB0Rl6FqsPPLghpiuFO66c8soZ0LrrkqQtDa2W6wus5EL9XG8XDrmmx QPf+edYhVgZ3pq+dEjzfp7H6oRNLcD638DPaCp2mtxTvaa9oLxcUKMcdv/axPZGI7MiHL1BtNdVZ /stw9WFZZx0vtxZQfZHK0oo7qUfVaoGryKnt+krANNqsJ2Nk4bRCwnGWSpWREKMpcZ1Gmz1tCXcX 9GTTTqMGdX5lXh3D4157LQTVQykh+deBswoN1V2ynDO14e3LAWpvA0zdTHi2ijOThv9BEtX3YBv0 trVDVFHRq0vDsrHWkPukoIp7XPUMIU99XuSrfd0IOP1kUjkSX8t7++Zjlx0p0gb7fk0LA2SRHnCv mb6pf9N3+62XLY+mYzeYKxwGnUFG7LeQJ68qOPbiYvx7NsEL/zTxZPCbWEBuNQuoHPMeMQToQS8d 0KPA2FDGJT11DlxL8kvSLMS9pYHoUj8ZX39Q/setdtCtYBfT4GMogTmv3eBn0Imf7WJIhtzRy4LF QpvrymWa/wEQFeoZQgGJlzwAWAKBXCPR5RyRRCJWIGtmcKIGoHiBsY2NA1XsQNZCQDUEOEJ7rtNX lRDQr7mAEG5ge5TuWNc7R4RJEStc4fv21rCH0a84gBsh+tTIkRyWxn88hMzHBoiEcyCQckpM4BGk 18DiRFACTpwA5jAQySlGMgGd6xwIMGlJyY2Ak2WJVQte94PWNIVZRLLflmpowrNRIWtLAeMP3rhC GNxgdgwbGi5pMcM8VUxONxwXaZj2x8UFcF05IMwCm0cErskyekrEiiybqEQFUNERlASAJK1J/s0u bhObV9QkJrcoORyCBxvoUdaVSOY8EZoPSXksmF8kt8FQeUiWkqPBPRZyTCfwjTZ19Gcu8afH3vGx f1Ea5lzUhYJjaqKBQWDgCpPwQGa8MQPWo2Q3j3NND1ySmx0FgBa7OIUubhGYoOQgKtBZj2QqsGS9 augzNZeyXi5JXG8shSjrac/H+aCF+nwhP2UoVNXlyXMX9KLOdJiWxCG0KQrFBx0RydKHPnMHDVxk xBw5TUhak6uRnGQ3E1RNbG4SpOBcIRcdkTR8EaQFPKuSXFhSSHUiEauzuCMtVMm6OtjzlQncKdRu pbcX9rOOf2tn6iZYr9YltY9Mbapci/kC/p4A6YBVteozK7cMzTaSotqEoDYrakWvjZasG/hoJkna xdRirqx+QmEwf1AAqXHjJdBrVjI6SzReJmixSAMsHwCrNUU4h1RAvYxQc7lLor5TI7xz2QYxtEPI tkVd15KKkbAKUSIWI4Kafch3tcrEiz7DehUsbUZZu1rWYrCkJo0uPVIaV2+8Yq6DhGZVUTkwEirp c8L9byvySRkYnqF+hgVoDAUKT8bOU4cjoa5TJTtZH13muxbeqni9urlsYtOin63gWE/LzbKytsTq BQA8tujaLeLrZYBCDwgIuI9/bSqBmCWi9JbgxM4+8qLk7WaHSWxNX9aUBP/dqTeKS5k5/gVVuQne sF5V9Ny1NlY8B4VwZKuEA4YwoaI3LoN3Iwot64ntw18dJxZbS+KQghQUaHXv9so5D1gWTxW5YkQZ jRjRG0vOqzvuakavCUU0+/bIbzyIgOeA3AL3zbDH0C8e+Ts4BgMGJIS4MpZLYt2Fwi+KmdOBZkE9 zYo6gbwkBosTP5Bq55a1zSYucXvd6162wrcgQNhyEohAFxqfDjHbHS+gw9xV81ozxBy+ZquH/Dk2 G5pqotOHkhXd5OQO9XxFlVa4MAgmQG0M05k+yKZ9xORFa46GvC1hLzNS5BUx6bUt5l5sR2lnGRtm P5mia6+260zu8liJ4Q2tFAON0SvW/qvZXssXouO0z4bID6B+Q/DQFFw3SuPU0g/+tqYlzACFGRcj vIRQb9HdXymvu6YYbPdr42y4j5zlrWHE7n3xDTHOBnvY09woqQOt6hWuueQGx5yTXJ7k9wHtltTW FnMlTmT6uLvK3cb4TzatcGl/vOpkE8bIaVpydveu6bRu6+FYM29k4luduc1sVWmOTR37W8xe8/CZ TUvyE//c2X2gczcGW5nCOrpEAVXspAmqsy4AghBQj7rGfQqGJXsc6xp1/PXSbdSt24tw2k75J79u llAaQOg2gOoq+u4stYO2w2TO6I9xPuLzakSTS6r7CoMuKnGXyg1OdjLuzr2y7NnN/umCODzie6p4 OaV7WMaHvOOzfrRlH9XyjUHbe+Hdh2ffesu2F70shM12JloW2KMGuDeLrexThzh9R+165qWPN5P8 VI4N3wr9IF47pWeb4qEEBN6BXw+o+YHjRZY8AO7O0lFe5alPhUDf/rybi03f4QTC2nAakFSAgdkJ 6W2UoL0dkFVSemUgav3S82Ge5rEcPZFD+41bBN7e/UBZcw2g/sAW5wGC/gXfKH3HHYBCAN6g8rHa hBDg4p3c5WFe5skZA8LMvuADdpkBO2WVhnXf9nWZsQkaskXhzilbFW3dD3rdSeWUSVAd32GfsyAW 9UTZxPXePHVBDEaYbHFPHlDD/tbp4NzxIOXpkw+qj0IAoQapnPo5SbmkhN48yJMtl9U9Xg6yYJHN YQbhYa3BIFsUHcMd3R/6odKZnP3Z2hmiIXr0SCooBA+6IScSIhzmhvMZoh0GYQgOoR88Vk9B4Jyg 4GaYW7RE3sjt4LqJIgguIPu0xbh1IR05Cvw9mRimzw/eTSVWl8bR0hXQoCaeHCh+oqSEogFeHgLW ogLeDXyJYBd4DyL4gSYeVyB+XMgNohvCoSFGY/rBmzW+Bu2wYgq6Yhq9IdOl3zAWRmC9wA1g4hrW 4OIxoz72ICs9Ix6QYya+mzmCHed5nhbEkQmeT3ZYmwpeGyzioBX24yHeoTmi/pT4zJEXFokvrmD9 2U08AlIxxsD0XYMe+Ik/jiNKnuQz1uEodgRFbs9AMuALdp6Y+F9YHF8ghtxDxqI7lpwcnuQoCmQ1 hl2HOCLuMWQknl9SfGRwzCM9HuOovJZK4uNPLmNKXuU/tuRLCiVB6mFBGOQhDN/CUF03Ht834uCk +aREAiQidiVN0sr8NBqqyB/SAV7+TAFTpktIxkAXjGScAWFWsuRUCuZkaGXKBaTK2YecbZ4pLs1b uoji/dRYouUg7mA48uBP0iI5bmVMnqPwGCUasaON2NBG5GWh7KVIQmWLpdxUtiZWol9LIuY0DqQu kMbhgGUi2KNk5qBZUqZv/u6jawLhNApheOCm02QkpKFRlJlmxzglX/ZfPZLkahomdVZnLW5lW9Km THpluXibmERmPi6ZZRJZJ44ncPLjVbKkcGanW+4h1KnjQgIi/jDnMDknDPSlatpjJlpnYE6kdSoH doagds7DC+bfJDgHVOCjDRIfOAqgLH5iZgZnbMJkRa6AI1AiU36hEqYgBtAnxtmnSBYEgcJbYk4n YlYBipakig4niyombQ6lCDbgHxjngR6jVPrFeTJjVbpmf2pliw6ok2SNd3ookRZpLKEmDXTBiKrh cP6nk65oiSLiYrYVY4aHFtZDC+1ny5wn8+noWvqgejYdRb6oHnqNkZ4p/prWAIjOQF9C54v+aJTC qZxOKZASaGM2oAOahF8iY0paJVL4KaDyKJgapmxSKJlaI02aaZou6pmu6QwQwOEsKYnOKaXGqZS+ KYzeKUhYi43uZzJuaXhC6J8K6gESaoliYh6KoNApKqO2al46apK2qYjKJKaiKk7sAa5Kp63WqvpV 6XYejmNmg4H0n3SqqEkWJqm+JmE66Y/uaqYWKFjek6tO6xnCag3g50jm50BaaqXeKq/2qp1yHp6i Ys+oprF+6msmq4Ru5mGeap16JY3+FZJSK70Kj7Veq6yGa7ZG57f2a4X6qr6KK7ASXiCQa3O4abF6 KmAW5rLiQn+y64TK/immbmeBYqMWsGq9ZixC3WsPQOqsBiy/+qvIgivF/iqwLo2BDmvI/qUdZmWy NiyzcuuhmixIGKx5DJfG5qy9zisiCADhnayblizIDq3QFi10Au1Z/IGlEYrHSmqUpuiTTmiKUqqz dibN4ukpYKzObm2t8KwiEEABXGjRQmU9li2Vku3Zmu3YIq0OOU3/XSiTsmzUzu3+BKiLzmyMzig3 aC3X9m1qcOwWOMLPsm3QEq3hmqLAnuzPEmy8ckgXsFG2Jiy3uiSU2i3VUqPVguzAPqY38K3ffq4M mgRKgK3SAmsoHS7qIi7hLm6e1icAFO6kTu7UUu6JTu5sYu672inK/jbuJEgr6P6u6HqtGwUQ2C5u 6RIu8iavjBpv2AqCzTqNI0Aqtkrqm9qu9erqyK4t0BIsW3gu8H7vNwivamhc8coq4Yqt8iIt8zog 4A5PZHxsyJrj9c7v7dIpyeru8hITT4Ev//au+B6CfQ5A+eKf0j4uADBvAScwweJfhtBA+/4t8Zgv xdbmyFZwraqt5qpvzTYlzvavBwPw/wZuCL+AAIMtAQguAy9wCptwWojwCENwrEbqBI9srlatBWPw 1e7u8xLj/n6wD99sxpHDA9/sC4Nkxx4v/A4tBZvtEqdt6rJc4gYr/qGL9/7wDw+xmhbx1mjxkbou EAyABAcsBVvw/gU7ccmyrbDubA9bsQ9jsQNzsYeE7s5uAemGcRKv7RMbLXdub2k0VRWzMfC68QwI siAD8f+sKSAg8eniMSOn7tGqb+tC2B8Dct8WcgEJMRwDgSVj8iSUbwIr7yI/ciin7/KCRBofnu9S 8u9uMvNwcveysltocfEm7yI7sh5GcdIy8A5j2SSrcr2yciHD8iALswplMgMIQiIbLykjb5uybjZE MnP2si+36iZbMjEPM60IswAn8/E2Mx8nbS4XcCAUgMWmqTRPM5pWszFjjRETyjXPQAmP8wCnMAOz sPNu7Tmjs4eqsz68Mzafpuv6c6blsz6/6jqz8/4d9MUq9CII/vTerrHOEnRBVyI/9zNDa7JDG3JA XzRFS/REA581K3RGB1F5jHTWQrTfevRHf1tIgxtHZ4FJxzFkdfD3qvRKz/RBVzQH9xBOxzRI0/RN x2AwM7RPXzIxURdQ969NB7XiDDXsvLQItzNSozT/LjVTU7Exw3JRkzQx8jJVe7BVX3XXurJLdy1U y7QkfzVYJ7VY/xEhX/RW/3OWDbRatzFbt3VTZ7JWnzUiXHNcn/RfM2Uq43Vedy5c83Vfv3RgF/Ni C/ZdE/ZYtw9HNzZaWzTUPfY0hzVkB7FkBy9WizRiZxxlRzNmb/ZRx/JcZ3NLX3Zd67Nmm3Zno7Zn Q29Wh3bw/o42kb42bJ/0Q6d2c+q1bd92cGesbu/28Pb2bBc2cqNya9/0YBt3cvO2b/+2bP80bi9q cUM3Rlc3Z6uxdOtfaeN1dmv3G3P3U7s1x143Yw/35443eVe2/8rxIf+veq/3ew9zc9/3QoevfM93 Q7N3ZNU3tbo3dAOuPwt4eY/vMIZ3gTO4fmfxd083D+V3K8ejgxs3gW92evM1gmcxkna4/4I4cV/4 g1c4Ixw4gPM3TKd4gLP4Bz93iSf4iXO4ixfzdju2iEc0iWv3hr8ySxdjjs94kOPzju/2vQr0kMu1 jFs4hcc41jQ5bB85Yic5fi/5Rxa5k2d4Zocwktd4+AYQ/pUPb5hXMpZftbU69JgruYkbdJqTOZSb ufiiuZcvd5sndp27+ZuvNKxm9J3LdJ8v9J/j+ZxP657bdqC38qFjdKJzrZaDb6HL46A/dKQj3qIL upELL59PeixrumhzunOXORs7aqZ3dKUT8YfkuZP7QKOT+XEnlKfH95qzOeakukWj+ot77Ugf+luU +hvPOq3XOq+rdmKf9k8bNY7P0q8/ta0HMs+adKBTybLntT0lu3AHu16CMLGzdq8LNYBRe4vvpbWX dVRLtbZDOHgLl7e/8ldHe6POa0z3OdQYyqsr+rRT6zLTp/eu+oJ/OHvDe0gij093s2OCs6m/0cAL PDgj/vzxKoLCe3PCOzxcNPzBO7zEx1fEV7zCu/r++jroOue7zzu237i83ntYLrPJHw5gnfzJz4DK tzxR0qPLxzyG2oPMm7yyowSytze/w/CHShbQGnzMh2jNq/xODX3ywrzRtzzNJ73SM73SJzSS6fyK 8/SPL33yAr3TZz3y2pPWH842eGzXK6/3DEDYp69OkH3ZK69L1zurizy5e/XEx73BUzzd+0EA1P3D x33e773SOm9aYT3eYzy21qwhQCoC8H3gJ/7Bn4X3GD6pKIDeC77eF84u6ATY8oHiI37kO3zKAjbb 6/jUU31amzzXp2/hFQI+pP2YHMyMDkAnYf0yW8CN/rB+523DAEDgVQx9hEjBbpDzMbMAQ0UVzSwk aJjAzjAA2ffhEmT9LVhB56/3528cGb9ouat6irc5+ipv0Z8s5fNGt1mfBPIiO9iBUuxMWpgAB+aY HY1/8e+BpRX+tTjEXBK/ynR/7f8+hcnl33DGURyM718+BByUFFXrLr1571cJqyqZEMQomJVt3Rd2 gZmu7dk49HzXD/7nAw6FxV4vllQumc3VjAl1TqlPQBWb1c4wHMbiGzYYbmUAJZEulRC7VGEwCEhC 3U0YnP9+uqCRmk1Cp6CAYOAgTcHsSu/Ow0+EJO3EzTCA4XDCwgOvM48P8i/wB46BQCdRxKvRc48D /gRDBPBkbMA05yTVTqOV9bFPdpRAq2nR2Ac5WXmZuZmZGJpKSmk6GrraOnsLoC7Dg8PYBm2N8mes kEAg4Gzz230DNhKNxKTNvFDgpGLxHR44ZNwke+fgXCJgQN8mb/1+hZo10IChW7ksdGPoLhRANeVK ESiASo3FhRf9yaOXYMcwbS7IhDNjJMiRmDOJyKxJ08hKndhe8NTpxOdPoTBmJFjVK4xLbuPquSFk KQCCOo6Qfnr1T1RTUoY0hTDT61s8UQIrEbiUSVeHqnhABSNLytYpRAo5+fLUVqMJcwECHJTQFZZa u43aMm1jYKgPpYucNXb8+NlQYkFZUJYcw/Ll/p8zLJLU4HLjqB3oBAhg19lzyVgaJ406x3UCiDIX xQLMao9UQUwIY09NraG2LHoQJR6kiPp3cKaBUjDoyyMR3d/AsRoeJBSnTR3htHfP/v1meB+as1jO TL7yFfSaZ/Q+6qtDuJPlCBpSJzUwr8Fhq288MbCQQ3Qpoyq83HItIjlWkCCt98C6yqHhRosLpEgc 3A/CSCQkhQEB/Douv8HuytA2shD7CbIUVVyRxQPWk0Y9omJ8MYnzaNzmAAuTw8AY/wQxhwD7AHBv JNqqywrBAOnY5AYjM5IkEBTeUJAB3kYIMTXlfATSlgJ4SwW5LFcbi7lhBPCSwYqw3NGkt1RA/tE7 8eKESU7w6LxzB1tujGJGGfrcswUbAYWRgXbWssofM1orJzf7ovJNRMIgHHM5+tCZyzcbHgSF0i3H CPISBtKEVL9DDXxoNAEY+AjTNQvEKJZKORyA1YRIjdQVTktkDjt9WvwV2MdSSGfQJYISFD1kizVW PQp2YZPHG5qCKEDTiJyOOodCo29YtFT5jAYnKb1NCEKo/PLb6YLblVtivQTxWTEjXFRKM98FLEzP lNvwTZ0qtMDOOQUOmGAgnApw2Rr/tCLhQBdumNBCIX1VV0XEcW2rdOZAjuJJ22Sj3VMaXKCGjrFi DUFLFgSsLpO1dLMAVT9ieaFDc81WuLea/qP1X45xbehAWvpd6Us1+1gg2KSZIacNaomFuKeFlSVv aqjTW2FkktbdhwaQnR7AWunkHbdSauEYNbBwGdoaVSnNDdVKsdnUEFUgZS7awiK1JpvpTzFBE981 9563TOzmAjOeoxNXoGA8Z5oHkIcoMRtsq/2U0fKqrZ4GvwtHVM2kricvSx1EArO5Ib61+lTkdsBh 5HOPgybOrJWz/vnm/nSOeVW0442UPz+s+2EYnlu91dTU2UVBJZ3iFulobL1ZXPExSySzHm7xsdxh zDd/mHvMYnRP+iPVaE97OOSgQ3BxP159SgbTBdf9nCU87O0V0L11cOvJtXtVeGvf2szn/jW/0UqA +eqfPHx0IjhFR0PUk2D1FteNCNINe0xLX+XCx7DLfa+DQBnfxDAUvJyd7z+jCZI6DDAyk2VLdSk7 iAvBADv4kGh2TqGS7wTjMt2NghAy813NfnYqfg3DIz0L0atypy29ONB5IIIedfQFOgr673qRY9qP XqOqEPJEc5IJY8Oqca0qPikgXjtYHAKAEB2NLYJlW2M+dKG3frDNR27TTZVsNcA7+q9un7qbFP0Y PNVBxBYDGAMhf0dANA5nDEPJASDydsUJXhKLGEyj5Jo2xxAyABtjFCP4PgnKPpHQh2JJ43++po40 PCt5oDsh/PBxuxeWJIdAqp2oWNZD/uDhsG6FuMTM6tiyG1ZseIUwBTGNRpVfnkoNO5AkIxNXPhhe UpMZZKVTnvbFGYlyKOAcFDZcB602zYdycoiKs+zoyBhucAC9VJf57jeIPe6vnX+c14Y+NUwBNrJ+ y0vVLf6COIC6k4ECiaRQJjkPDVkSohK04AXtNwsNUs6L3uwe98S5J544y5mxMxDKtnkPIdFwU0Bj V5LSYctn4iyhLA0VDz0nKVmS1B7C7B3NaipSwnFkAAAAXDGJmFJocmSagYveQX9xTUteEHKRmxY3 Q6XRq3GUlFZtQefyWbGPCSR9UJHnGQEJpQ3SUW7yolseCXKuPjL1HWj0lE5nxtPy/v0UQM75kFLv ekihSRIRgYMeJgmbyXNqEaydvMcePzmNju4kqx386BJLKDucas9R5CtVZSmIJMzS0WdGBaanhjXT sW42lblU5k6LGdJjyi6ZzrmXQV1rVC2Vo3k6KQA1l7ojp0r0sGZVo0mrqtXHruS46/FJ5/p62C0u 1lGmgyusfiqaN3DldgusaJTs6dbedJW6ZOOnAWwhAFxIJa1kfV9ePXReg4LXkDFNCWBjo6bBRjSi E6XoH4SbPcoV16rJ1YaA2bOwCcDyhk0sK73muE5UvtaIgRwWaIsUy2vO0mmmdWkqO+Umew3Rc2MQ 8YhJXGITjxheFebsBk4s4h/o/uknu1Uq9ar4287GkUwMph2AJaseAlvjx6OEgXTh+9u1JvZgK5xD a92RAid7CcopOKHOpiQgsUUZy0/Wcpa5LOL88RG9YerymLdcZih3yL1pJfOazdzmKBdyAW6WM5u9 RIy6Iu6+hcXmfvmLWDX+t5QeDPI1IgvCGJTzlvti8GLZ2JsWPxrSJMZsPCkQaUtfmsQadh2mOW3p cnYa1JhGbahJXeIpyPi9vRWTjSsYXIuOTpc87vGgCR3ojcIgAXSec5kX7eUg6RrYu4byAZ4ibGMH O8voOvaykW0AEDAb2sjWQLSpTWYmpMm+esYvJAy73UVxl9G2tgKtJ1NoiFWj/tTpRlC62d1udz96 Au+Wt7PjPG973zvdSeBtNffGasVBNcdbJA6MbU0DrQYaB81WuJvxV22HLxzi0Rbawyn+ZkJUHONt jPjG2xyDgjq0ktre836jqsVtOi2jBSd3uUspYhzgG+Yxl7m75RDvmYNaAAuQy80xDbaD8JzTHp+x qmu87QVrU7GfIjjCV76NDmp5BhznOAAwLnXynqnq1Q7QkFBAiqwrHB8KGCoFrI5s/MnBI7gou8OF LliRb1u/FC05OcD9mqUzvbHmplGJXw70EqvT75B+SjpiE/gRr3GdRTP8It1mlgXUyq4w52fM2ou3 Sgc+p0JPNb+N5O+tZVGq/nVPkLiv2mPulTnqX59ztdYR77UDWwjdEsDxFKB6KD83B08JKi0+bpHX p8A/lpKD2HFBSVWoOLUCxd8CPAS4rqiZ4obBX9sh+NC3j1zuAfcvFydM+nEfHGJScjnVgR77AKkT P1ORfK8zRuEHv8DUKvXs1xKfYge9AAwrQBRMzdrJYTEf8q4EwToB/3wqViZvAQbg545DbggwCY5J ePhF84yG81bN6HBs7rRChe6ulAwO/BrGAlIg4VSvpJIsHcLmjXBH/qJqqkwqDmivp2wKmPpPAnyt 5rrO/moLwvhPjuACAC0vX1AHmhLrU5gviXqPAmPwc1SJn3KrBRLo+C4w/r+eBAP7S/S2h/RK5gMT hp1IBgDAoNT4BsmCoK3QD6X0QP9YwKYU7bkGISIIL8yKCg3DIA3X8J3+K95yELXyjw43Ba+4SQGM Rw9ZIQ3rcAk7zDUKIAEX8PmCcA75EMIi0ESSAG3yzAL1LJs2SeA20PugoOmw4BOdgEm+cA/9sIB0 zARLg+tSsBRl8N9I6grVBwZ9iWJepg2Jbet4bxBb8RC1ZQxTYAB+ENuaSQmpC8MqAQAhr/p+RwhH y02SAAitTwqxD+A0cfvMr5tUzpS2cFleyWIcsAUiccE2EbrAZskQbQ7D0QC/SjR0KZ7STw5hABIV jB03qP7sCrUKsA6H/pC7/k8Q43AA01ENe/EYSSEBnW8Z41EJ4MMXeWUCQ266+MPzTqbP+gs3toID A0zvaCQUm6BzuIAX9w+XdgUVoUtIluT9VmwcySKdZjEk2WIFabDruiUqcBAfVdCEVooj4OAHGdAR LeyH2uUgeYC3aPGZ/pDYoFGKwuX6KIgGqDATt0X0VqgTfWwjX6QjmSARQJIQ+xACgWGV2rE+zLFV PKcQe/EOuQ8f4KW2zjIPDAEuFVAuB0An8TAB/ike59EQEUUBDIEu/+c1ApFVjC8gB3Iv6TGXIgIA 0q73orAtvZIgcao5HlIEnpLo4ko1mPIVKwpKFq1RUg7vPAirLAe9/rgSJy+MJMmxPhwFITaMs2yx BcvQJXWwE1xCALblInWP69LsJ48SAwiABgaAk7orARkvIY1JtMQQQeQAgXTAJ12liJxRGJSyEZ+y KamxCk1uuMpC1ozrKpXrOy8jNrgAHOURJilSjsxGyU5DIV1AHOsx6bpFH+anPN1SKehSwt4mD2+y PiHTA4JqBvCTrf7vCI/zMdXRDmOKI8ziCIsyHx/wPJ2IDSZzyALHMouMDybykVhwuHKg+7Lwm8Kz wLgnATijGOkREWPzuvySL+yqGSMs+NTTWxxTCBejbTw0QGqSNykLKIEDQAFAOOsJF4NRkaKRGaNz MwMJDgSgORsz/rSQlAcbKCPBDEw0k3pq4AI0kykv1Cus8w9qYCNoAAFqQJdAM3OUInyycglKlBQf dB4T9ABfLT4HrzQCoLVawS0jtIBUNEByEE/9swb8IQ6Cs9em5B6ZzC7ylDAu4EcFlMrgoC8HMyR4 VC9tK06fa0kbdOge8031lIHoQwlaU00u9FlmwzK9kBtQtQvMoEtvIA0W4brM9HuMYRvPlERNlDZd cZ/oJaxYdJ3K6SX5Mihj8QW/C/lEBEv/gQZuUwPVwUN2zyZTgQBUjAB8QSIgBDgDVHLA5kx4skgd NFdzBzhmKRGNcAxQYVNP1LJoEIrgzyets3q2lDOe8kK91Dqf/pJN2UF0aqANaIB1uvPcwqFWbfV7 imohPTUt1TOz0LEPEfQ8ebD/Jo2mujIGULVN/YBQN+JHaaBF7/JHA3EGqjVbrwBAc2tjgbQDGlUS ThYAAiAYk2hTkaIAEzSHmqNbnZOn2rNiPRVJpAkaMcUCSPV12tQLwUVtspQLLIYbJqprRqBrxPQE aGBIZRVqBFY0q1ZEw2lpkfM9U7RDRSxI2OgcO+NFZ3A4y7E1BTAv7yJQlXUGbnMSWBZqDcA0AjQ4 EzA4g/MLRtYGgtFHCbUC5LZlixMJe7MWgxJAFtE4J/VJgfJ9JrRdXQDVKsJK+2ARZMNEMTdVvTQS 8JVrhgRq/g/AX98QYMmIVq82YM+0PTtVJAnnF1fUHM3ITQeSZutSLQ0BxEJKUdu2CwaVM1rDGG6T vPI2ZHWOBkZWJYwhGBk1OElAef0CaBtXUR8WLNGJQ7AOZ2nLLP2TZ2PrZxGHVL2BHyw2VZH2C5U2 I5qWAp62KKI24RgrTU8XdRNGTcWnAV+Kns7WQ6usdBD1RdmQTwPkn4yyFe6zNeoWAGYvZviVbsuA SB0YE/TWeEMWBEZWOBMAgWuwJYCUEVMtXEVyXG3XIBXQONP1g3OFCXdHCSbXaTljccI3WrihXl2Y cyvzd1+1PULXX9GhdE2XQG4tdbGWbPlwZ2sXe1KIDNEh/naloz/fFEZf11wmtj/D0SUIICB+9B7A dHjvNgP4luAkeG/VBm8DNHDFdAcQ+EO0tyuJmHZ5Ni8Kp0ntT2fNE2iGxwklN3rR93fUJlz6OGlV 1XxbVZDzFYe5bkzfl0q86YdLL4gDFlhPM0ktsgfGkoX8F38hVn+vS5Hqq3HtwoA1Nh9Gg4ER+C97 F4xrqHj3oFExoFHTAIFLYytybq88WF1vigaJrVpJuHChs0cPab6SgACW0gxW1RiI+Xxn+AY+VxwO metE9+UMIWuTZZGB2HRT1xtn9zBjUk6dolvWh6vuj3uFFT7VUn348xFptwbiEpoztinAxlW3GEgD A3n1/gB51fl4mZeME0Bun/Xxis+hCnN6W/dSnygiXBZNDocYRSSgR5JcIhePq298N6BtT9UyKxdL hQNMzwd9mHlKmO4GvAdrU/ebP3ikjphRylEdhiSh//cUv1b38sGc1yJZIQFMTUBw26OB39YiGhUP +NYMlteC1eCmdW5HO9lS12tBFxMhGZeXfXO9rmMJRiXbpnEKSU77thMXJaIDp5mal6V+L+cM6Rhh U7Ok9tdc0G+IG1ascQgW4WmsplgNedcPfhQQbgCBzwCep9UO5pkFfLoMqnWM41mof3q2VvqcZ5Yt UjhKaqHynvNI1TqbC0NCHNoF8rh6rGkKj65sNsij/rUQpKuZftUDHmsZNZGOnNdTdiEZk2M0Fd9q bVmhbSGBrl15WUmhZHI6nmGBp9/yeEGYlTNWIFTRBmY5CQn4KyURU9FuqdWWa49ata7N7ah6z7oN ciySPjBS5Tz7s706mrNBCsjHAVkXMclaLLu5RVMwvNtYnC8VisvZkiGzDuU6FjKamWfvHG5bZfe6 t+nweA9RZetGuBPYSFdXvRFzeMagNDp4UgvzsCOz/yiUEjl5gooOE10NNzm7szPjqz+oWKahIprb t10Xq1dTFY2VtCN5m00Spt8bWcMXZEUnX2cPF2/7lXf6lE1pMRshDvT6xYG0kGV8DNZhBtrISSk1 /vki1m06RFOrz3Dfs2bv+AWkOoKk+6n4bO54dSCyekpn9Tw2vKs5ckYYlo0j+8LC8qQp+VenAq73 kqFFeCzLUncfcDEuOMbN4a6r5Ed5XAP4uq/3e4Lb9L/59YxpoLDRWyAR+xOQcmcUnLkPtMAfYXYg 3OOYvAIpvMo1G50wfKs9kFkaGVCqIaxfU7xKksStZaSL0auQvAfodBBtxkY1+m3vGgpwu5TxGUg9 QXn/VlsxOByW2zGDdax79jDsRRn/+ViPHFUou7IbMQqv8xUzqbrNHIltMLvNzcvnF8xbQMzDebzJ FdbQHM6x2RSRu71xl8UNEw+Ugin22eDgmS71/juV+VBwAdvWBfSmsQ5c05vM9zSvdNmxXxvdaVZW lL0F/oK5Kx0zNRTT5TR9qHZQbqIGiEBhQPvT+8Q1RWq1SR12TQMlAV5XlRN+hgUOz92TFyEOKmVj A6AAhBxIcVsAbDxAW0EB+zZbWlkNZL1lO6TYWVG1EVeD7YUoTThYTahsFiqq8xginR0qnYvhs1z3 0ENppBYyZgAZxqm7WW6r3g/Ru32V5pT1wv3Q3TPY3VyTyzID1tzBu7402sqL8JJkHlQd5dLA0YlJ zwToGSnOixiFfZlDGDPfKfbRGVpnmiBtkxBDr8Lz4q7PLhzlfqJOHt8IqH5gcgAH7gQ8rd4F/jSr 53eVtU2SLL8LxNFSQcWSNHIX2LU5jzYQbhqTR04YI7zdaf4me4v79Pf+qYHkb4Kell2fH2mB4FlA ypudymsD2iuy6ZPM4adAaRxD8lOk+R1DyDzqT3LkdNZ43BM2FdchrAN67jNZ98wenMUbtqyX0eDG /kjxT1lX8G95WHbD7Ql835FSmJbc2MM/8DPEe5fgeSa83y4wE60bAtA5xhQimN68e0ZN4WSIJWmm 6EoBwAifsRur44fnnqv3fs7zJBSKhfHIWCSXyqNzUSRKFYlqAnGVSCiWC2EwCASww6iTiW4+o9Kh +5pFVQjfA5kIPTfTzDW0TQWHsDVRcBGm/lEyWIWn4KLHp+bERvRWNUhiQcAwUDCxWIb3FJnmN2UV xxXWeQI6lTe6J5v0ZHQaOChh8TPBGCqVZyYcTDxMOYXsFoiKlbtVYTiQ8eMhYn2Nna2d/bK9TeOd TU0dNG4OBPAxBKnkMVt7bJmFyWVIJyAQwFiE1J7zXgzQpVQU7A04sA8WGh2yjsRblgpaAWkaCriK 4gLALHelJv150+zZRE4VLooysvBfH1t/KlmhdyEAAU927gDrx4djQ5aVcG3Z5MMAmV/Eahn1w7NY S2Q95YWkBw3DuQ01WMi4agPrCgBaq8IA17XrVHRjy2oo1wFBI5TvOq5JBtEZigKawIix/lmL1M6A LgfSy3TITkJ+bPXS8sjGqRYShwIkqpkwY+G2khwuBRnRwBcGM3vZXIuT8mHLty49M4CPQAXBn83E Moz4o08S1EqgemVMaW5jjZhS+a2M2bzT0aZNpRkuufLl2sAxd2x2R7ro59ByCMVug7+9wgQKWlzP izQAWE5uXKIBIOI2LzMbRFgG1mTt9LkHA/4d8EToQvE6epQTem495JdIA5AEWXyEhUZfegNe9tIW EwnQWYK4Zeegg3vxhB8mBtR2kW5HGUXYbnwJhF+BUGli3Dk0hZVVjFbJ6JVWzsE4AUXUcWDdjj30 qIFaN82nHmkQqnjaHF+EEQB8oBHp/haHTeESx2l0HNRadpSZ4l1I1hjEn4VFZKQRlBve11SEmWxW oUkLmvlgPGoWxACFqwm2z5OinSmnaYXU5uQva5lI6Im+AQcRkvrpeByeVDAHaaTNdbNNVND5yACQ mEqXQ0LncVCklCnmF96V+ZTnWoCgxtklQV3QYcBg8zG0Eprs+RmejhbhRWY6e+hUK4eYQTWSamLK 9+kH9p0yZwF10vVJlgxm2AFABMbx4Q/95TnMiCOWKOKhUwonVz0EXOoiQkOFMmO7NdKIYwguaEXc hJtmOt29ZOGAkGtwVrZemqb9las0TXqqZSRGMlugHYDV4V9ee5oiGzPgzYEIA4rY/tTrnsHaOqyV B1YIB1GvTRxwXCJxotq2Vlx4sl4LdyghoLcBU2jOaC5VGqLk5rJoi+YQ4KiCShkhadLWBOJCw/Vo gu6Omup7Vr7qmJfSdqwKPJwc4p1q9Hla83kZlVAVdA+qWP8qYJHdhbwofyH2qlHW9Y12rZcHRMPZ xrIyqLWGtfaJ7SasRCvrgnYLDvCRWeyirZtHe2tZUoTCFVx75SoZ9Th0lfwbMuGOvpu446LSdLlx 3zu1vq0zgLDHlXXHtd4lSPRFPrGG/e9oombuJXG5Ozmk7L6DbPbFjT12Ed29w1O20xhzRtNQej6P N8jNOjtAy66YnHCo2uMKVFCC/oV+oc45m+4zKt+tGE3n1BB9B1H+vuktuKUfmnkCqWdmKddZjWob eF18kkWtxh2jL6R6mF3soKeUrGpZrcKElTCAKoWwDQcUZJh77MWAXe2DbnXzxwQ/lpjkMWZkfuMd Aqt1vBQ2TDOcgBZ8/jYrlczOcYOg4Q9EaD9kUc4jljMU5uJSpYfJ74exWlf6SAdFnPkmFP3zH3mq dLsuZGxTr8OUAQ/4PBJFT3OEKJXBAIBD48WmNPmplzQiFrMtWW5Um9NMxlr4n17BJoZjtJ0d+4Yn boVPYZOooIQMBy1QCFJxaiRNl3ThuRsGUX2k6w1TgkPHJOZqiT6gny8GVRQh/rKkiPvjGYpQd0X4 WWCLXhwgAbsoJAy1LU4Vi9DZomKXMYBmcYR0ZO2egTYMCmlts2Rc9mRIKonoyGVEIKGvirk1msGk ABkwVogYyUtrac8ZcwjA4ZoROlFmM5q4ytYPJdetIcpHfUd0Ste4QAdO9kA1oBMUJdlpuio2wwVA C4FE8iFAAnYAlmHb4xq907B63SMAV7weyt7CRj8JDwzEI+ZDf6dCkXwhTK5wpuy4lCY/SsVYgXKo QWcGt72dayZ4dCH2oOdBD0XSeqK7ZyXzmcmnuFGeOvCkMnC2TnWO8qZTpGPqkhS/gAq0gK70QCxx QqsdXjKji2LSMLFpwrtJ/tWQDoMnBl0oIA4OzpQWC1pFOurMsGqVjyFNhSGqiUd/vXCCXBqWKuzE TLBuUFljZRYW/jQOdIpSnf8h6lTdacE1lW8q9HQiUG1aSv6dUnP8zMQ/X9lULma2A/3CH2woFlNN 4g4DAmho8U46x7YmtmDEE+dH56iYFW2UeWRwZglllrKUziEDibwZslAL0Q5haz9typJnX+vIkEGO GmICpU1dQDtTjiunF9OiWXyKvkJBj5S6sWQ7KUse/WyGal2MTnlhh9UG9fJtSCQY7gyGF8CtFaYI Fa1mhPmKac2SbKp14zQ8cQfb3jaBfX2kh97KGcTpdZwNyZtIUlOS+kVQ/mzz9eVs9mYORz12sEfh xyPW1zPEVne8Y2nsJ2saxSiyT5/fAcveuiC0Vi4VX9RYsBwtDDz72oO05LmqxHqZW5XBbyIHEaR8 xZc30S5PYwkS8GeDLL0L8FYRJ3btk5E3Q2dRqHoSlutLUUolxmTYTc5NMXSNeNj2vhOX1w3kJCtH 2IWhebJ+saJ4D0TezUpNzx1YBz/UCqyD9rer781Ha/V7QqnWUqIOHAQVfpvVQM+MgUPWFSgEXLdA s1W4MDlXCHvxMrlKUL0KrCWxWNbSCQMahlsFyU/MIVj8qfMRhv1uQteE5xI38WZlhuzOJHu6OiOA BhWISox9dN6xJDuW/um9Me3aW90djweCp0UuRi1W6SK7tJGpFbLIOIoFTN8WyX3MjKd7G8QjR8kh BkYANJ7F5UUOEsmPpM05Yp1ijGgkupLF5M8IrYnoqKbLvR6s/m4q3fYJm9ivOvael5psTzEYhWNk tBkNHbsAkTuia77vABxdbWhunNOE0OJ/8SRukRc4pIuBmgC4HM5mk3rlNBPzN42bMAI3uI+QnIq8 vZy/BZwZxKeENqGdJfBd//SJvn52v9XsjHl1QcoCTbY5rF4eSOMWx3Ts52gNJhTsxPFj17alQrGU 7jBCWU0a9TSTFyHui2I5mReoSF5NilwHr6mG8f6plR+qdwyXxZ75/g7Gh+ccYs2J1kcD57WJhHpw 3jydXFmQuj0c/vA8wxqML3RbuVc72nygkfOjNiZ9FQPwQ4RIvprudllDUJyzhjvlra+rClVaEVDH XJaJ3rSrxfzp5op6rzqxvZ/MaZZ0CtUItL7c5Ml4gHvNRJFvbvqvixpsQcRB6iyqOp/LYnXYkT7v c9epZT1u1Yxbe+7Se5W28X7lZ2P727QVNw9uXHbbjSQRglU3f281HHOwAa2Ac/5HdjxUM5jSdIdH CQnnb+1FQAMQdqFWcLIWeeHiXYknbBIwL5fnfZp3DuM3c//3S0N2D/nwNxq3btcyMBOlYbI0NhT3 Sw5jXbk3e3E3/oLHVG+6UHfX4Xcyp3M6SGnP4FQFCCVBmFsRslz3snxjAgD4VHRVMGM1RH1MUXhE 93wbCBUuYA88ZRbh9wNgGDv/QiKYY3ZzURfpx3lkSF8leHZht3sG2IZdd3SLBQKDYH/PNHKDtjdI JwRxOG+bxnIToANGKIdCWHPIN4UiMnSIBwhT+AES6FgFB3ly5nS2pn30wIUghFlKZQ4iGGnrdR+x hVSmwgAZhGgVth6k+DBBon6lpzDyNycl4Id9locDNHLK9QPDt2q5iAvjoHywmIsvAYnk8H3FOA6d UIUbZn0NCGwKl4mn4QLxBHHHOA5gKH4WtXXyZ3T+pEV41lp//rZ+LJhMS4hexfNlrneG0dcDCkIm wPU7ZjcWgVhqDBMdPyZ3fYKM12iN+9gDkuh4o7Nd+bZiOZUZLsAoIChjU8FsRwhk7FeO9tBnu0Rh jGNhQ0gPOaCNSOh7KjOPZFKR5GQFTKhyOzRjIbks/jgV2KiSHKCMk8iMV+iI2dcMHVcA0+iFX9iP YbiT12EebAhSXedeisgBqoZ/EOkM5iBzJJgie0ZrVyYqLSmVytaTU0ljVjkVAEmBjxdn3LVAmFiT 8OMCmJd5yFaVG7CRrNZXi0Yq45BDvXd6y+AjxBeLeZBnLImVeXl1ZymVeDmVQPRzMil5DugzUFcv LpCT4MeX/h/glwyQKuMYPWZDHZA5BXo5he9omZnpiZqJA43ZkrbxMkajb09oFJJRKBlRkIj1QZ6p A6x5lTvSFqYXlD3BmbUpQK5pm5aJmzO2m8gImITnhPtWDKhJOqhZVKpZR/aQEdVolgJ1lBySm9F5 l4spnZrZmx9YnR/gCQEpDJLRlUphmuIiD4qSK9fJVM2Znempnj6AmevpnueZneY5hT4VmKYpmM0E Xe2UKNF4Z/L5muZFne8poC3ZngOqnv6pWQO6naHpXMRZC6aJmiAZnPj5hG1AJoa5KPgSoJ25oQhq oB9qlh4KogtZnSJKNfQJnELHgIZ3ZrYFBRGanxRKJtFo/oJ4Jp+NaaIjqqN7maM7qpjp2aObsqCB mUdH4Z0SSpx0MwWoaUWjR56X9Z+sE6BB6qNVypjLaaW2SaUruaFSiaJPZJ/d2aKHZ5y9cgoZMZ7v c3ZM1aUDRZ1bmqVxWqBxqpttypwKKkm7N5qyBqFDl6QXuqRK+jND1iLX6ZdwSqdVOqeJipWIyo8G +qUoFqYsSpr22aeAaqG9QqMKVah2Cp88WQ6Oyqgjuqij2pee2ol4yp2UaqS05p15FJwqWqHJgKbQ Vy9Cg6Ukyp4kZKq9ilmi6qvAuquQ6mZguqJ7Kqt5UKb5yaRomkdkZIIx1pt005q25avXGqKoiq17 WaLa/lqM29llw+lRyQqrLzqmFXqhJNSNDyOtb0p7HIpp2yqvPOqt8/oj9aqr7hmpF+Ki5WqpH2ac sJpHTrqp4nVshxqv6JCw9sqw8IqvDcsp0SmsZUFPWykileiVCcc1t2YuDper1bGwV+pkEEuyPPKx JYuduTmxJVasoimYfEGYVfSkJrcvVGmtCnuzKDuidHABPUivOsubD0sdK8tYSqenk0NEBDl5ZhMR 5sJJeBmy0pGzQJub0HI7/gQvWFtsXWCyaNEJ9lCL58CzEgmxRBuxA9p4W1mB94lTyOl1NKuwiplW OEutVKuZdHG1MpANhFBG2pC3XGAAZKK1g5u3xUYX/kTzt4mrtVvLtYlqtm76oRVLpBwmRhnbtu4T PEp0r3LLq3SrR3ZrlVcrAovhDM6QiWpauqTbVRNAJtCguq+LDYmrCA6jupXijWELoo9rsh8KkOHa jDC7tO6jpqtTrTrZuQ77uaDrj3diDVoQliUjvNH7vDolFxyYEVq7K6mbWH27uIaQurQLvimQRe/2 obr7qQJqYj94sUihtMfJaU0bQJtrs3WLvCervEuFHLRbk+6DKDg1XdEbjYJbRlkkbMNbu3JgAaf7 veAhB1wwoOZrvtSgldyytpCFfdAYlm+bVLt6lnNbvxE8qp7QCpAhYfbEPsG7n/uUEQy8uAOAuajL /sKFWwAvbH4x7I3uCcFCO2OSW32UO5DOd5w0XHJ0sR/GOL+lyqZIfL87kr/fw3TLt0ZG1KSYxsKZ 8L+Kt73WcDsEwGI1HLs+JJ05DKK9W2W/y2/uC8CZu0lGfMR8Rr9LvCkj4MRAB8Xsuz9SkIfV28AW QJMGrMX1gMF+vLhhrMNUCaLpe7TKB2dQmGaUtyJ1kZgayqVKXDWUDMfnQIBG9lIGdTxRPLACJshy gImKMsB87G9YbMMVoKWFzKVjPIEnVsFXeMFdrGPUBLJHfK/2e8mNYiGgyMlyB1PiimkbuLreqIEF O8Bc0MfmJ77mmJdiHLkzbLG5gbG1xnHtR8S5/sbGvCACvRIOgosNuzxTcLRqsAmPyjrM1AV6FGAr G9tGf2wAYNk1pZyZ0GygZKy+ZtwSMbufTqMfZMmhA5i1rJsROCK48XIV4qwBCiYK5TyXDwlT4hbI emzM8szMCHzMXmwCqvzMrMytYwxEiRxKScvIFv1BJMZcB3C9kALO3+AcSkOIoEtlB6SSKHM8El1f 5AkYZDXRQANM8ZzR86zFROmPugvCPyCB6gLL+nyJPbHMWGS4UtEDVXEjA73SYUFsA/0uKFCyfqM4 kLh+0INpdFawGw3UZD1/4Xs7BZB9pIzAuIuMRu3RAmUR04y0lHOBTZ3TEflHOeANL70cK+3S/jCd HA2re3Q5lRclYDzdxf1kxe3c00P5lWmqxrczlY971D+k1Pn8smfcyDRMEB3r1+/yAgidAvMiI1yh 1aa9AvM60yiR2FBZmrb12YN6McqMxqg8lGwd1BSdRQQ61zypoyHtsoo8VCXd2yeNeYT915TC3DAt r/1BBA4d288pq84z2ZF92wUA2Y29zgbA2EJczBy9j5gd3FQDrkvd2dfXyAb5yN2HA6y91TPgLpQi 3/IN0FkKhwqQl6/VyU5A27ktvBQNGAJe1l6T3KsFGOV93vILosRNid8JxIcl3qWy3GTw3Nfg3Bm+ HHWBrfAB27UZg+LjTJIXheL9092t3UNd/tuhTAFx3eA1++CbLdLNmIFoPbywBzU5cHfuct+rDeQr IDz5vaO28WeZGdZr4FFf2dbQut0G7tYV/dSqA7jFaLaZfU6+u74krWJZyJ8W7tcEJwwcTub+7H69 auShaJnc9qBK2l0hhqE/feMyK8gVQOG2TWgvfpkxHtDDTeM9bMFMXphuCxj3xUnMhGKF4uNBDo06 LdqjmuZ6OeIr8d9JS0Iai+NeDEy8feJaeNsV8ICO7NhVHrScieU+AOFcOUR5DdnXHJFUxy84By51 HFR33N5PijbaTKf9op7ADOAdY+K3jsp5ftbCHuV2boZ4PsClbp18LqR/Hsv3dONNzswS/hETPD7H 6+18/EzLoHdfRP6h/aLm1b11q37dMcrto+LWxBGFhvndsTLoul3KRJ2g9ezsmAJgZbzli9zlBp7B ZsUvJbVhQ2XulqvidO5HHpiovM6Zk74Tlf4EwN7qnS7vec7pVyzqkh3vyj4Xd6qXp47q0B6TgZ7u hG5Gh05mTM3e4qnuOa5YkfyeQlEEki7bQWcMZsrypzzgda4KOO7pxA7vaRrlC56qH3/vmIJdxe3D doyB/VbhysRJGgZKBN+Eljjxy+x1fZiQVRp2497f59xh123r+rnzGo3scF72pavFBIDxTg5wdpiv Vgny8yTy2g68UJ7j77VZd6fyg4nC/hzvVTC/nggR4uRu3WFvIuHZto1e8T8d9Iz/82bN+Bms4A6s kB3to0m/tj/c7y1e7TUY3wI/9UvP5ZFF9m6fK+DunhDk9TUN9pUrzMF++mUv1CZA8RlfymyN8PxJ z5vZqEePKS4zuST/96j/XjqQ6tGutJhO7e/EGNSYpQxfjA6/bkFn8BSampC/qZUPDdpP+T491EKP zPB8+b9vpVpu3N9i9YLuzgkP1xwAmk9E9dU84RatwZzoo9JP8/5t/bpxrDkPASpJmuxFKW90+gEN Y1Kq6rI2Teu+cDTRdEXqNjwMZud7/wcyAIBg0XhEDpFLZtPpW01ICwXVWsVetVlu/olaApekEhmn 9ioUCMdCzUIBb+Xc+dYbDo/1GNY5lzMgGAh4KjQ83MlYWEB0YnxkgJSMXKy0vLyk07wDwMPTO5GS mTG7+TAABB0l7TPFGRFFUSn1eNVp/FHC3RUi4v0FhqurwiQetvMUC6WxCWEqOHgjubNaNMYsrtS8 uvuM4bM5NUgbBDY/9Kg6R5psj3zHjtc+rurESlZWJRudDb/5MxBjXxlm4Wr9KbCMFS0XgM7pWucE YkSKTaZtwzismyd9slqBcELAg7Qw9Ex2uYcv30BSzUAcSCPIV0WaP2AsiuhOEk552DJuUmmPo8BV BEudaYhKBAFpq/oZTJrKaUFX/i9FFHg4s6aRiVu9AtnW05Kxnypj0UhgCBpJOtbExiOrEd+eln6u Fij3dasiRl+H/AWgk+fbaie/qATTcR9LtAxfpiJqlOrBEAWKLmyWNJC5rnp7dPYcesfPkxu9nbXA y80YPIZJpzx8+lvdUznUDBBd0QACL+vcAQZMeJ61vxhNz00cGTW4xpkpA0p4ufHHf2gkP60O8NYv 0KG7586dze1rswrWRWNNLe4l4H8Ll02ZnO5k6HjBR9ydAKe5vvB6EWEkOOHmcY+8+IZaSZXLZmkF KVuW8wg7B7UrikGoHlOKM63A++4+D4NYYIg4SshtpPReC7G94pCJDbnlaEHo/rYPzUmHGEN0gocn 9yrZcTh6VkQMMcUgvI6qCW1Rg5/p/MlOqXGYw66WqETQcMYOZ8Ryhyu/GkmYLNzCpscURZzHOBcH gvEu3LLcJRobf8FRwEcwkXNMBVQMDIvgSihORG8SnI0xC6l77iqFrnNOM6YiNLLJHNbgbkPRtmTT SklFMxGOkjIq0IpOUWpRvtnoE2EcQSrFpUaegOgPx3bCrJNHFevBc8VZ7wSOqEArXJLJql54cioJ fwVknEF9dXQpYCiliVlUObw0NBMv0gJMS8Rc5NM9DxvTHgVpQ8O+Z9GxwBL/dmiVB1dznDNb4Ngj s8D2xjysOFw7aS8PXedT/nJYB2u7Sthe/0USygaJVQqrSD90dlzvovVsWmpJ+5QKW2+9N9cXP6rv VIcL6bKwdSuhRDiycrWYTFoDs5dPP+8NQ8S/hmxKumN9lRLD+hZqlGCr6rMQ52THqfK+hj/W6+jz MvBSDlgDgxfqi/O0E+ZR0zRVAKSfQM8Lkkf2r90BAVv5ij5VrrrbXIubZshDGQ36wsegW7A5uTF8 8tiDC7WtaGi3RlXpczK9SD05KoYZyMRf7ggcl+7yGPAlMv1yQJPd7cTTIVLO3GV8X7Y6Y+ASmFk5 xvq125+cp1xK76MQLraNnolF0u/cBJecItyBye+NEd+jc/Oo6dVTZtAB/o3w8axzX2I33uKwPHri OS+7+M+FCr1lZdje93SB/YVdBLhT99kWpuJWvaFXIF3YaIiZf9hDiTc9BvGzhbJT+6tJHUcNreE3 AiqYNoWMjMc1nLMH2Ty3wNDdawxsm1npLuC98aHvQtkBVt7Il6jnFOx15ftZqbbTvtu9D4BJM2FN PMAbAjoteFCTVea0ZTzs6essWPPfCY1AOcNdzlqawJPZbqW9XInuft2TTgUbtTqdiW9JSwxffiwY RYUtK4UV2Z0OrXgfELCFftuY4eaq1zJ7ce+GHFPT/7Rok9Vo6ncoAtWI8uUJBTKQAoB5oARnlkQl go+JU6Ib+C7YxEAK/o2JIbxKVty3Rg9lERfO810PRYY5FWkjiJ4b3f4MAh0CqJGRPABBG994QAOF Kkh/QqLp+MgzQbqCdSNwXfpyFkLbfJCDrKsiCSd1xU9GxJGNQAUKvlgHPIFRf5jUJBr7NwBP9hIa K/TiKaU5TU21LZWxWMwqDTZFJv2RkKVqgxuQ5U1ainCcrvwm0RT5t16WEDzOa1pbohcXMyVjPhw4 yFU62c4eCFCYhYPNiDbCLYFSszzIQ2LNkhjLc86yiYF4Yjdppx3LIItg6tOOL3m5jl/ysxAdNUQw T2Q4OMrFlPtyXD6X2cxeimCFGZhgegBq0GqKym10yaY2i8RNWYav/lgabOhEnXQXiR7yFT9ln+08 A1KPNoGpT4DkxIDnQ3ocRzn8y2FTGRBKFSiGptK05k0TSiQlsbIgBxskRp2UhlYaFZEisCg657ZW 3W1UqVrdylObINJIlhSOVh3KGXGIF5Y6k6ur+SdJxLrYXSlUp1GI6MAsSs65AeSnbJ2sXNNZgED0 tIOWFWEa6upOvKIwN1ElqeZg+JaToe1AskmeStVACLyiApow/Wf3lIFQxi7Wsat03eyKikHtLEW4 mi2nsYL62aGOcJ3xK61f7Mq7rgFUiJ07oL1O2riWyBYvtK3tYRl0mWyGIqfYjOljp2LW4Mb1op89 qgivklm3xhec/iK5Jd+ay9k1aZS00fWKXpcQzr5Wi16sLROQCApbNLFApQQgAHi16lIpdXUU5e1t Y8mqXty2F62GnKgt5rtcoRY3nO8l7n6bNV1dAnjFolkLTIcJM1ISL0j3ZIZ381JaCot3OuoF8noh 62EzfJjE9U1uaAWQVviC9rLQQG46Q1sTAS+hyi7OBYt3QWCpZmFH7okV2RSsQDKKKLbimK2LKdwC G6AlyIl9M6/YG1Fxuve9lX2rqY6c4uIaYMkoZu5lC3vXvGoZy4a4chGgPFIuTG1eK5vjrPQ4hDOH a8fR7XGFwxHZGXR6yHzwdIcj2946x/Wl3iSufUMLZUDjWSl9/m4dklV93wAbGhGJPvRnbH0IAohS kl8GNtRoTLaq3Y90lG7whJaSVABnGllP9PSbRT1nIhs5fYBWK2jBeWJ0pjrJnaXsXPss2lpDN9dU 3rUhoGld61WPeGPeHBE7cew9Yka2kcOyS23rSn/0YdT/JnWR7bzncH9z3P0ruKthPQ5wY9Tg8u2f aZea7nNbmeJP8LUxpSa8A/docRiTYLL/sZT+HtrZmu53tVUOcLtZm+AuyParV81kb8+61wm37bhz WWhzV3y0Ecu4eqamWuqp1ogQxKNRRr5Mn+u7smy2s7/bPPVSS73UAz8ntpv88MuaKuGVlbl8OXtz hzt5v6bq/nm5ff5ivfBQnsGOIeau20AwRBDZlbYNvnPt9LmZwr1SZ/lTrI51iWo9lHgup6BRjfj4 /jTWNTc7rdPO87Vj8eJNWLcn5j69TsnL8wmkoQPtjeZ9rp2zTs/5S7r9UsK3HuqnXv3qGd/45q76 5oefvdm7noYAKJz2Yh/HoNE98cqznUsyHmXZDrz8WkEa5JTGO8mLzwBovFr1f3R99gkP+8WH+Gez Hrtmy/77u2xd3GFPJDulO33L6wW1rWk358WIwLRlzHNSqHeOSS/8Q5+e74iPPex7vczQugCEudz7 A9pzvG0zPwVEP4ZTA8STuYWDnEa6PCbAteLLwB5wOwO7/gaxoKc5KA8c84PK0LvKEyHLwjMDRIoW HEAX7L4YHL/zi7zQwq8GTDyI46xnyr0J1MGdUz/KYz+OukAkQK0uq7E4GsFl6Id7479zS4MUtD4E PEADlMHrQ0CF80EKhMAInL0tfEDbS0Ddg7irgEBmW6T1G0L/8ooOlCfLCUGgsCdBcbBTkJE13IHT U8HZs8I+vMIDzMKcA8Mnw0EcID8GZLxBhDg0tECJw0NCw4+g86vSeC3kUTonfEQeiMJSIcMs9MRP TMQc1Lba27ZOksBOLMMuFAkH5MIdBEJLccRM5IUN3AE3dCEEMyA5DKwTaEKruENZpD4pdDJQxELK mEFQ/hREMkzF/um9U2RFHewfAfg+VAS+AnjCIDQ+YJzFIiyCI9S8SQSKk7LE0XsBCLvG6dtEKUS9 TjTEMUzGdkxGRazBZXTFVUTFUazG/ukkMNS3MtxBZvwYWuwFbdyirbDFyoHDMhFBF7lEOyw9gtwB AUjHQWxHYvzEd7xHUtxBexS3Z+zHVYNHZfxBRgwcbiwCgdRAk2Qj5JuxJOSGSryqFHAOTpIwiNwB AphIiMvInZRH9FtAaEwDNehJfExFVxyAGvxJCNRHknwWgUTJytvAgzQwXKyqMxGFNMm7c5RFnPQ/ +eLJr+zJj/TH/hlKfizKCFREsTRDVwQgp1TJTNxA/m+EP5cMqDn8luZ4sAizySMYgCj8x2VMxWIR TE4ETMD8x3okAB9Uy8I8zKBMTFI0zC5UAy1yy73cxpoQqZaUnjhEBrtsSBCICa3cS78EzMEEy8Gk R1UcgLIsrqSsR2osyv6ZzE+ixaeEyrfUREmky4XcrrchFECAsJq0TCToy6BUSjMsTMYEJ6UMSjV4 zOSMzLVcSgGATulszBNcow20zdvEzK5JvknqCc5ECbvkRTSCCcIaTkQYAAgjzb9kuOhkTtl0zuqs zsaMCfoMLVVkStrETR7YTu6sCC5zo9QCR8BCKXIEzggDjPTcBQhjT65szghtTpyEsAEoTuM8zPjM /tANjdAK1dAPdcWY+EUX087+hMv+zMwWQsgBEc+X9MzYqozvqhUTZVAmWE8IxVC/lFDHFAT23FHZ lM8JrdDpK9EabQRcE9C5BEfetEqncIly1Ev/nNH/NNIqZcNstFKJwM21QMIVDU+F1EUGA5f6uLQs m1IazdI0vczhU1MtrYhekzF2282CEtNkq0MnkYknOFM0bdM+TQIapdK1w7XqSq3NBNPODKxLjNGS u7Uz9dNHVcP2g1SLqwi3I1C6NFB+wYAnzTtGjZQpndRQLcifE9WTxE1CLSDCaFGYtMTB8tS6AtVS lVVKldRZBYIrI5xvnNNMvSqsNJVXrTVHtdVh/sW1QE1JikDVW2TRQx3PRIVRNWGTPR3WUC1WPj3R imDJ71zSutzFCuFUtEMaaZ3WLK3Wcc2yimCNOqAq+CDP5LELdYIfcTXXvUw0YyVSIlhQBnjQ2TyE Ad1WFrFKJlQmj5LXecXDerXWNeSsHFARsJtG9HvFHlDWdZWLF3WcQ9vThDXYaMVNewUwPVSKGQXE JuPHiJ1TgG1XpRvCgt3Y6ELYUJ3CWthTgaMvPks/H/jAL81FRIWt1LBJlm3ZthTWRgUOP/0/lDvT 74GiQLsLI/ArXpWCPgXaoA3XWCXaWqlRHqwwyNKDKe1W33wd5nIuqj3SoSXbSjFbRLNam6y+/sND LEY7ujoSErD9Tfvaz7PV06nF24nD2rKdUbZ9jLfVzBiaFUq0KZE7koza26xY28VFt769WhWByDWL 02HqiVphVnG0KSj5VoCIWMfdBb0FXb/FE1xoXDxUPeQTmbB5B3SBB8xlF5/g2QT5zBAa3RU73dvN W8k1XcjFw2fKOHOZBNcl3teVXAQLU9rFjBJUXN193L91XovjXdJtj0fkLFGqhtal3s1hXbHJhmbt 1d/MkOiVrtyNXt+N3L94RK7yHXP51HcJGxB0URvynkFqXvLlW+h13tIN3emdvsx7E40qWont3pKR XfAlwc79XPwNVvNtWf7lARxdYF2r3ukT/iCmcV/LC47uld/kTaZNAhYG5lgHHtYCUBHWPDsC8N+1 66IJIJkGhghXGQy44FkFiRKrGFsRxhLRVdP/a48/nL0VPjd/0g9IQCH3KGDknd1RAReAyWEdRtu0 ZdCjPbz2cD2HAoEfBqchrgEScZ+uAJsvBV/lTanamGAojmISZj/rg7rxGh2dQp8fRqQzXqP88OIv DowkNmAa9mBN7S7NoGM0bkoeVrPUYzOuNQGNuSZhAQ4spqumQo/9YJgY1mPvzYSKHUe8BBhB7iVC bql9WzfF+iKU8eBxxKOiAq27BRx/kmTw6A/u3QlL/t4lpl9v9RlUCGROZh5PlpyF3dqY/mpJ66Ga cGzSpHOMOdaiLlKHKPYBGZblzH2tu/zjV9BlAOPlStm3yn1DPVbfZ57lAwk5d20SeJWcYFrmLHmE PH6Vyynlze2XmcTlaj63a8aUX6aGS6jkd4DlPT7gzjQjxB2a3HGTj0kXwVDidr7LJtxkeRZUenYq SvHlLp6Cr9Heh1aXJJaHRquhBAZhgNgaN+mP+4gTftZZTG7XQzlmNGDoNXToIBhgIABe1aXoQnhl deZgMf4C0PHjur3fShlohwljVTXcbkWUfIKJlQbGlhZiAF4VzigCZ6aTMOXoKUGV3Thnka7kGc5o aDbQWhbYlAYEYEXqg5ViH/hb9v0S/t0h4HXe6nqYX3dOnYfKkvwI6Rkp6Jsm6X5m1699546mkrFm 0IJN2okOj4s2aL0OFRumjxfAkmhoZb2468M+6IUEq5jkXOZ1EsAm14ydUSP2ipHp4CXW1GN+ZPBI h7r2kDCOXZze6zrllbvJAc1+VM6uYLvGaD4WbTpUttg27Tv2DND2ZsQm5huzbFaaSZWW7dmmbfX9 CuBua5T1anE2anLmkgn4kMiO3+AeC1JaQmmOa81IbuVebuYWDahm7d7cmMT9a73gi8/Oau2+ZPjQ XL6mIMnSrycO7xod79r2jfcW6j6mW7nJ7OO7avd2XeemWAOZw0zua3jm7fxOU9q2/u0Dv2nZ3dm5 bXC8qe4A5g/Dvm2hvnCTxnAN2ylZ0g5VhnCIhGBX9vBYzuvtbm3e4txCwW9zaG+awG7J/m/XqGzu +h4OwuEUz1K6Rm02ceYXB556YvDFdgiasGoORwQEX9YQV8JpuqmFOiv7RW4hN9IuLvLy9u/Jbmcy lm4npomReOwOB24kx+2/OiWaee2lDXIur9EbH5egFvO3jm6lw+waT5XnSfMlyPHsniceZ9WT/pb6 laxbLhYUp/M1PG0skfKSNhP0ZmLH2W3qBgaQ/vImSJfjZRc29xE3n2+bCuf6niJH1vRHB8ZIr5RB h+9vpmXLDprpdnSuAXR3oHDg/sZc4R5qarry/FNaW1L1VWf1R0SFAv/tFq9wGqZyPXftH1dvY0+V cnlhQedmJQgQbNlxNwf2GAjnurkZRr/ZY2916371fP6abhfxw0UdAX/wTcdgG1FteRCTTnn24f6q K58AYe8jEHsoajd3C+aNTq+JNU9I+X7z4v5uW1iaiV4VPDfge+c4Ulat6/GyMvscTPKWjwMA/EM2 f6czVK6dgZdFRZhwdRf1fH9JzfXuMtdwGvFO8Cz0A8sYBFKc5js6jv94FQF5n6e3oGcvgD8/cDV5 1C14yH7vlYdmyl74l98gGvfzQoAn6OkJpu/1d1MZRxsjGjoiB8qYfjez9hj2/qVlnaPPxAPw7cKe dPll+UOH821ylFxuni5R0anCe422eL0HPez5egkIPVUgG6EX+jceec+KOYFHey4uYvc2b7dPcuju cRKHGyAfcHkvsMrJ2byfHo2bmpjx+rVBtu3JJLYpfGOW++XKs8WHdAQI9P7N113/8HVVsDF3d5SO eoCheyNsIzl182MCfYxvoPvJl3+2u48n/D2y4iwH+MQvd9avPE4Hhm2fl5Jh+h+qKuBHDhm3GTkv +V9g3y7NrvmjgzLyE3kT/t0KOaT7eNE7/Z6HPobKr7lab+ivPNl5faKV0n1+fDGvfWiHACVnUvVa mzbnKH0hMh7IcR6GwbCt/vvCcXGOWzbhubJMPA4AdRMgQEIcBhPEohJYOR45xM20CZA6rSBiCDT6 kkziEiplVqli6jW77X7D4/I5vW6/4/N6VkmyWNwBsgAuEBEyGP4pLjIu7vztRPJMRvodCeVgaGZ0 dHqAgIqEkaGg7bkYnIBiYFa6+uxcCi0tIdEGKdw6TWXoXmlROSFwWQF0gSGPjZWdGRCcQkdLT1NX W1+3HVT87QkiGjIcii8t0pYDFRZN1vJAydoy5QY5cWp4bhx7gZGmqgxEq6DxgUM9Vq1sSaiQsNe8 JBZqRQmWBBiWK1WIDQPygctGjclGLSvTLwWaAthOokypciVLOX0ehVvj/k1cTJrmGun6oyvSrXY7 fcXbWSzLPXyhQn0hY6IUyQImowlcVfDGwapWFRbktWkrp6Je84HV91EpKZFNn7ZMq3Yt27ZvtMH0 RocmoJ7c0gHQiS7REbx9DckrUiswQsK3NH09moxsKQNoT80o8cXTVK6WK2OOUuWC184DFYOe/BEZ 2ZBnSKpw6nY169aurb1cVHM23Zp3cZLTm3edOp6A+xrBlUOWZnpPLBJNDHas6ZLTUkle1QHz5erU fRW00bno57DLR5dmbBaNs9fmz6NP/w1IC20K79KOX5vQbd27gSq6FQud7yI6iOMCFGEEeaYYacqc 0I9j1EAnhmid2RMh/oHaUciZJ0ts10l3oYVWw2j7ILgUM6eRV0B56qGYooop+TKQH4yEU18jM+J1 Tl6B6VfjJe+8w1BQSQBF0SdGfXdgWc2o9lyDD95DnZPWIWZhhvds6B1SojAXnohMlbiil1+CqQdQ Qcg2n3wzJqKbmpDwduNf/AXn33DC8RJFPVXYQ2WHi22Z4IlK0iCZh0dNWaihHWxIJIfHePhhGCCJ x8xI5BkwpqWXYrpEmJtyOk2mu9EYqiP2+cRfjYYFmKqccbJqp6tA7pJclQbyeWSCTV0TkFIPfnao r79KZ6V3jiYVoq2Solbpp8sy26yzz0ILbadfNiuqtZD4Qkm22BH2/oMuWjxkHHabIfoJo+D12Qw2 ugaazKKEBjtkvMK+e65YxBqL4IjIKljiM25FG7DAAxNcsMHSXhOtJAtT0nB/l8CCo7eCqYrVBbUc 9yMTFf0CzLBYgnhsvwuuC51AYLwLrKGz0vvdvR9qeex4JT02rc0tHJyzzjvznMgjk7DJsNAOO+yH VTpYTMGTEiq3Z5Zb9ltzNQE1uKu7LdebtbCN4ltrvvvym2xqUt9cttl49BwtLGsvbLTbRx+dNJRM F8rylS9Dmu7IKrFb2kdaA4711l0Xm3ekI06abAH/nt24468xmwPRQwv9dsRw4yD30hbmWaCiRXo9 Hq58j8RMiF3r/pO6aFzjTXjhXxsLNpckluQY44/jnrtamE5nEOZxK73Q3FHSPaXdorQecrq3Kpgk 6WagcPrVgwpeL+uuvx6zvmAnXqIAuoMfvkpALakdVb+jX5Xmw3clpefWo3u4ui2ZSPWI+T6Kvf7o Gv617MyTSGyO+ZP4CmhAadDiBeXrlZMW4kCqrG9z2eHcoY63KGLJD4DkuR1L6ge9dvnNQfsbIf9g J57lzW5kqSHgAVvoQjsAYA26GhSBpvLApN0weBKEUoVWpjWYxU50qWlL/ewnOxMiUXtKNOH/wtY9 AZrIeS+cIhX3MEMm9c587NtKhYpXQXP9EGRPyyDtUsNBtRjx/n5LtJqgBFXCNiYxiCi8lQaheMYq 4jGPcWhXFzqxwz8Or4u/siDojLQ9fjmGbGpxiv1KBzYRxXGN/pOZ7Ez2RErVzil31CMnOykD6KGs jxCaEAV7qDLuBO5KXYvZviZFM/N48In7gqQkaynHJrYygCocoBQ96ctfoqIMoZSOV+hmTFKe0iif C2PyRBieXI7se+dhJBp0+UFcYjOblUyhLjGZycVJE5jiFGfpRMireCWzXMpMpb1W5yg4PvN/rhzb JltTxMTRMZu0DBQ/yzJHbFrSmph0SjjHadBfmsyNHEpnOgkZRgxOEppoqOdrqEkegZ4mn9pEVj4v 2U1vfpOi/gcdKR7LaTVksDOl1Wun69YIzTIOUKTmEYBFPerIhG40lx3F6C5LZDuSApWcoDypSuWV qKK6bH9K3OY8WfglAtQUk6jhKVVtClOQ+lSTMg0qV6dYTdO50VGqo94FkefOZr6zWEuV50cTudUU QfWeVQ3gTuto1aYONJFd3atQE6qMDyE1sO5saRKZiiSn2mwAjIwiVik11WpClmqSbWxWebm4t/I1 sy6MpTBDGFYSghaI/Tsh97g50AIEAHxQ5SVIHztZ1/ZLgCBlrCb/odnbApOz0dPSCMcqxtCONqKI u6peD6hYp9CWtZT1ZhGTS1vkEgCzuJ2uVyH7yCXmD47Z/s2eWiO5T7baFYo/reJxF1u75eb1uVql LntJetFZ8tacwMUXPCMJUOIqUo/Rje7ikOvf/0L3sgOwbXsLzFXOOhK+3rVlLft53+4xN5EENjCF K7xXAjQSlwvesEspOVy7UiqK+bUwiUucW7nic6PfXfFuczpcm1JWkyaeMY0PiuJrXheEHN7ed5sY 0EuGuAAFrTGRi6zHqFrVxUrm6Fx7alkCDNnIUp7yCzF84xTjdMku/jFPQ1xcKoM5zC6k6QAda927 NjnNeI2wiCcs5jfDOXxIpmpdPxjQOqN5zSKWbpz77GdOrfaeUj2zmj8KW6w+l89/XjSjn8pYbxY6 z81Yoe6eG23pS+MurspFL6cb21zLajXKmB41qW3G30d3OtXphW6pW+3qxw0gruZ1Lmtp7VzoEsDN r941rx0Xa1n/N5EA1uRlRd3rYyM72cpeNrOb7exnQzva0p42tatt7WtjO9va3ja3u+3tb4M73OIe N7nLbe5zozvd6l43u9vt7nfDO97ynje9623ve+M73/reN7/77e9/AzzgAh84wQsO7wgAADs= ------=_NextPart_000_0000_01C85294.643E2190 Content-Type: image/gif Content-Transfer-Encoding: base64 Content-Location: http://infolab.stanford.edu/~backrub/repos.gif R0lGODdhNwLOAPAAAAAAAP///ywAAAAANwLOAAAC/oyPqcvtD6OctNqLs968+w+G4kiW5omm6sq2 7gvH8kzX9o3n+s73/g8MCofEovGITCqXzKbzCY1Kp9Sq9YrNarfcrvcLDovH5LL5jE6r1+y2+w2P y+f0uv2Oz+v3/L7/DxgoOEhYaBgAkKi4CCCk2MeYyBDJyEY5Sfl4GKNJkln502kgevPZiAD6YBph eora6lpCqtAJC7FI0Upru+kyC/L7m0MqPLM6mqn6eYuLTCyZ0GxSjFir25DMbP0c69zdm0LNISzO U/5Cva38rc0uXc0OP50u/r57Wm7v/Rp/Dv7hDwM5aEUCsijGzYFBZ9gIyluwUMIlTPUc3uO3rmGs /oHx/o2ImMvigVn65KWKhvAkQ2/c+kGTFAymSJQdE2qsUPImxJnjImnkuVJn0J1AiVJUaFHUI5Us a8Zs1IypSXuaRJb0STQlz0ogJ3Dkp/LYyIlgqRIkmw3j0plrOw6lCVeoxKJG68raCrVoxY166V60 C/hhU7NYy7bkZfhd28R8C491PLib2I9sHUZVevlsVcwvOcfdB08yVpIy3YbOifEn6sc4K5vu8FTw X6SNabd2R9cmaNmZG+Pj/Htz0t6PpREPvRu5ctDGhfvVAKs26+KpeVu+Ln360JY0s7NKG/ge5Nlz J9/F/ba6bdnh1y3LqF63Z+3WvXPXrm/+/O3Y/qsDnzcZaa4IyJ5ge31GIH30GbTNN1IhSI+DaCFG EVna7KJghqq9JmAq0cG34HDBgYhffwommKBaufUlYV4tAihehurAleJ4b80Yn2vBvUbiYLetx5iG BTL23I2aucTjaQd21xySQPIn2ohPBrUfeyhahaWLHMVGY5YwzoYYgSkqKSKELwopHGxARaQViG2e h5J6QpIXoZtr3hmlnGlOaaCJRp7ZYTbgRUbkYp9R5iSZHno5ZnF+Uuklmp0l2eOh3zllGo7kiVAl l5W+KRSom1b50JL+ParbdIFaqGSFRy7TaAhfDWnpVGdWeOKRdOLZ1aa1VkpdqHhuuNqhFBab/p5W bmn66596TvqpriHmGVes7tEm5VHNejBrPoxGaqW0Jd5Ka6shvWhtgXXimt6c5j15bJHptrubqJaS uqdckNpXZrXgBsZsvuH1eqm24Q64X6f9Hpwjta7+CNbBiQbLLpOSFinQiuTyGSFuzFK866N98kty w92d6nGXDsvX739w2tUkws4Fq1iDufqWXLnL0VsvjoQth+yiQotHacaZirgstHM66qi4TK93EsuT 4ryzwlFOrWHMQNvc86vMIQsdXr5F/V7TWTp2mNkbj6ox2mWv9Da2PjUYN6I9LvluVnO/WjfRLj47 9qoL18ck2UgWFnfi4OUNkNhEX/Qgaqx2/nhye4Bf+pSn5nqFuMyDvqyvqZuX15zaDCrtrGE/AYZv aYM/XrjHOUnFFcY2EAwE7h7tzqnt4OjuA/C8D99T0b/7ToTwxC9vNPMGQ/Gg89J7gvwmyuMA9vTa F799995/D3744o9Pfvnmn49++uqvz3777r8Pf/zyz09//fbfj3/++u/Pf//+0xCdAApwgAQsoAEP iMAEKnCBDGygAx8IwQhKcIIUlOAwKojBDGpwgxzsoAc/CMIQipCDFzSCvIx3l7n8ryc1qB4VXJjC giQJhixEhgpXGLYWonAnTKAhZUw4wx1yC2E3xGHGdKimHgoxhskLIieI+B0jZsCHQ0zi/hKoKKsl PrGIvoAiM6R4RABqcSxKLCHp+IIJG5aLdk70VTXodZmmvXGNY8zdVfKTLTzmbI6ve5rFnle5vuyR JQ97RZx+ZaMhYBEYQRyNS2w4HlbN8YuJiRNk9iZH11XGCpO75CU8Obo30g1pNfMkUzDZkFCSyW9o nBZhrtdFM5JoMzzcmpkOScl9RUyN+xINxTb5QnFpimv1kZnqRjYtKI0LZUlzGqqEacxl4jIJi2xc FA/5jFwZcptkvCYvv5nNb4ZLm9OMwnC6ic5JilNk7PSlOPkooXSqs4/P4iY43QlPfL6zmlvEHo9W hEssyTON3gSmOtVoUF4KtJxPECSG/tapT4Oe86EOJSg20ZPKh1oUoOEsJz9h8NEaumdi1tDoc044 0EkuNKEqjSdDnVBRbk40nRLVZzfZGJNasvSlOvJiIe0pz5DG0p9nXOg8XXpQiyrDdJqhKEbRadSU NiGmULXpdWp50GusspJM/Q1Wv0rTCY30ap0sI1HPNdOrBvSfbVRUSS8K1qYyVKhIzOhaPdrKuxaq RT3t6juTClSFvhWw2wROWa9Yxyw2z5Azrao3l/rXuXpVo4KlbGTNqlTHAlWtepUqn1R1NLDyNFuV RUpYuYgEuk5xiedsbGkhm0u/rPSprx2oao2RNKy6lrMytSlq4Wpay452nz7VrVxz/knNxDLyscR1 7VHxykWjStdhtSVuFdIaUat657mStW5Sj6vZpXV0cNg9lWVvqwL0WuCk1E0WUpEZWKUmDKnErO53 lbsDnzlTu7u0TsggOllp6nKQv4SmnwYr2O2qFwULDgnmIvdeAOuRuYt7LyqpE9XLTtWUZoEuGUHJ Ss8JjcOvlIlpSfw3s22Wb5RLrizbcSfhno22wa1cYJHW3PsK15x3JI2HEerjih1zyDYOZG53VuA/ 2vZq+pLhi63IxFIot8FcoLKUeWdl5oqUev186QWynAUwixHL+K2iNU8gZiOdGRBplkGbw/BmvY1w znSus53vjOc863nP2VsBn/8M/uhAC3rQhC60BnUQ5y9TKtEaju//GM3oKHOyzOldNKU55+gdP/rS it7yWbEQaQf/ttKZ9rL/IF3HUJuax0DENDryClspotrTVw4zp9FsaZDCusayvnVraH07Xz95pI7u DBzv6Go5ARlTk3Ukz1R9Rr051XPTdik9TyNtQO7SobPrdkdzXDhEC/vTMKalTrVEygzzuqyRrF1T IMlGGYYFxZr0JShHSVbDHmbGeTnxKdtNtoaVEpZQvsIJpQPM0gm40b2NZs/MBGuFr1rcDh+m16Dk cHPpF+EH3i+1V8cfk0EqmWrO77hrjdxlrxWfu801ROsZVN5+m+Hkbuk0Ww7z/pKH96b0ze62xhnz FfOcv6V288mDPWqhe5e9rq6pSWnLW+8GgarVbazT42tsu4oWnltPaUU56lzuilsLB3fcyIEba8he vbdfjbqOHXHkzr7c68VNeI93ynVN/7Wvjf5324/eaVtzzu/3dKqWM6rV6f79xhEOXtzZrvS9110s nTScoGg8XLV59t17LXqXDY7CeZ8W7bxWO+V7vvgl633sWt95Vndd9XfhlMWgwDvdk9x30mqVsGMW /GI/7Nuspz24skU9Zd2u7h5Q3eaRByx4ed9k0A4f+nZneWgLn3KUgz6HjA1+gEtPfA4Z/+awT745 Hu96qz817Nkv7PVXbfen/sMWwa3X/qR/e7buzn3zcrdv1DHDeKvHeum3a+rnc/o3W/LXf+tCgAXo U/QndsMWTKgVK9OVcdC3cuZFfh+HfQGoSAYWIj+2NeVnIi2zMh3HgVkhYKVUIhkYgiZHdo2USOqG HRP2WBV2Xu5GM40HbeWWb0R3WiAWOyIWM/OGYtSHbvlWTADITK80gNs3eCkhY2TVdaJlg0smfHDD g4DHSJkTYeq3R7wiZ34jZNt2ZHcXczP3emZXc9d1abfVgxiYbG4Qh7gGCVyIW70zD0ZXhb8GB3W4 h3wAiIq1XIGoa761WnEwiFwmiHhodIYGiZEoiZNIiZXIOFJgiZmoiZvI/omdSEJgBIqhKIqjSIql aIqniIqpqIqryIqt6IqvCIuxKIuzSIu1aIu3iIu5qIu7yIu96Iu/CIzBKIzDSIzFaIzHiAfz5YjI yIwXlB0E14zRKI3TSI3VaI3XiI3ZqI3byI3d6I3fCI7hKI7jSI7laI7niI6HsHtd2AKLOHGBsHzQ s4zuuGHrqCaphlJzOFTHA36sxn8FVwiaM0T4yFZ+qGseEY8NBXs/pI5sOJDc03796GfLKAb0eBBH Z5FKMCuNQ5ARKZGk9g8ZCZIMRpFV5pA2AnD9lTOSxHj7VhaGp2w7CGPZdlEpuGzudG2zhynYomFy dQ5ZOCRYQ5MvWZOG/lVtG1WSW7CRg8KUF0coPtJ6adE5oodvZNg2+8ZXuHBvAVeVQaJ5XBVuX5eE S9WVwIdHRhhwm8Nu3faPe+ApxHQcWliDyVERxiJzHpcvythMK7gwTdJfozRghCOY8DUucSdxrpJH c6lMgdkn9mRxWgl1SWlriYeCJvMzJBeVFuaAOdeBPPN6Qdd8R2V9QJiAC6hdhsl+OndPB+h6zNeZ osmZcqgHFFIX3mKEYAJQnvd8sXduVbh8/8ea7pd5Cel8xdWTxklVayecMKl1u+maoNmWebAQjiR+ Z4mb70dISGlcw4l5m5dQOHd7n1ltstOdzimWqxd25vl9Q4cXkNd//oRJc9JZdrVXfGgplbkJlq35 nG4FlJf1m5vpmBiVbiOGlZ0neTVmfhImVszZcB2mn+qpegHJXntim1VDmdipZPa1n9dge0iYetBZ nAoKcHzHn8bpoVIXV6fXm1PIhBoKoSgqCLbJcdW5mJ+lm5ylnBjYof8Zmm6Xd9EZk1rYfucJf4jI nb4JXi9KWCYmgG5Zn/5CoyN4IR8ZoOEpnvXHezwKohHIc+/Yh/25Y0RaaqkZnh36nEranmbaiFHa NSInpSiTdlEFnB7onp5ZXj2aXdY3hUs3fqMnXsiJfm8HdPuneH76mioJpHcgo9mpHCxYmAWGoeT1 jMhXmQQWmAED/oRvel8c5y+3hJmrsZtUp5ef+nwIVqn+2Z4SylZl05RU41ZEKV+tSoSIKKtMSiyx k4PkNzo9hW+0N6BZ2W92laDF6TaqZCFvo5hb5Z3dyWbziXH/YqGZA34oWYONBzJmOJOwI4Lc5aAq SIbh5q3eKqbLCnLuJza8ildjg57MSosiaUcIKZlr5qWw6K6OB68b1oftGq/yeK+IlVm6WK/n16+p lVO/GLAwGJL7GjaPlI4N67APC7ERK7ETS7EVa7EXi7EZq7Eby7Ed67EfC7IhK7Lm6Ikla7Ini7Ip q2ckqbIt67IvC7MxO0AkGXjyuqICq49r9lHuurNcGGPoo17V/tOzpmevOWtmsilqH0iI4eAU6RO0 OzS04Ve0SSdSUau0hci0WCo+T1uzExmfB2m0UGa1cLe0NPuv48O16zWP6ymBWlu1lMazcOuzTQu0 Jzeh6xk9ooetSNsx5Mku0Jp7Cih8rnFuRsqeP2VkbrSphdukewpkqPquSMaoiftzn8eIchOsUImr YQWZDMeSlXeWa5GfOop65raGHdY5ZPlvqyu6KraBqJuWSetYptu4eUhvVhk4Cut5wMZMC0eYBsa2 nsqXGQeX+zV8LFZbj+lFbfOCNeqXOqe8hop/RAi5oaBfbvq8OrOPkpaDk/djpQqmeGqaK+a9/GeB cmeApAcw/gAKYFsKvjmHPOebae44XsHZmLvrtXaIpWDqnMupvpl3ov/7o3VqhYD6oAZMul9qovzb Sjm6n1oWqnQ7tfO7wMEbaml7tgxsvhYcvFaqpfGngPUXwQccofiJdYfZpcfHwRGKtHSaoPSLeYty s/ibtfqbwSscuCGKhKwaqAPMngsqwgjcUjzMoFtXrRCYTzhYoiwcv31avRNcVak7non0amZ7wwea wyCMmDK8vzhMmccrxL4KgeHrwYNVfUr8xe+ZfSN8thIolA+cxCwJtoZoxDicwDpMnMc5wylMpYJa qHJYml0MvQJakIDsxBAsxLXLhxhSv3uMsHQcpnacm1qc/q8B7L+X3MZMPH7FV8SRLHWUjHjiF8I5 y8aV3HuWdMg+/Mjcm67uq6fte8dVSsDAGcaRWsqUWsJH88rNe3ay7MepHKelrMhPhJPU1ZfWOses 3LuM+bsVR7usc6qPCakaiKBj/IAd18k2tnGT6nENR836GVvW7LZSJi31VbzbtciQPIZHTJ532bmG PKIFGpa5u5DzrMlD6IFgV2Ip9m7snKexC85j9cA/a69uA5mBI5rRU8U2zJOny2yn2WwyqBpDubeI yreDm8pX+LjhpyNpyLbIVmSfzFoYrXcwHIb0F0chXVfqPHW6e7RO1sfcl855aL2X29JW3ET8egQm HEbzbcrSCw3FamvK9seQMA1TLm2QmTx4Pv3T24uzq2W4T30eMkvVVW3VV62JLIvVW83VXe3Vczay YS3WY03WZW3WZ43Waa3Wa83Wbe3Wbw3XcS3Xc03XdW3Xd43Xea3Xe83Xfe3Xfw3YgS3Yg40+BQAA Ow== ------=_NextPart_000_0000_01C85294.643E2190 Content-Type: image/gif Content-Transfer-Encoding: base64 Content-Location: http://infolab.stanford.edu/~backrub/barrels.gif R0lGODdhTQJMAvAAAAAAAP///ywAAAAATQJMAgAC/oyPqcvtD6OctNqLs968+w+G4kiW5omm6sq2 7gvH8kzX9o3n+s73/g8MCofEovGITCqXzKbzCY1Kp9Sq9YrNarfcrvcLDovH5LL5jE6r1+y2+w2P y+f0uv2Oz+v3/L7/DxgoOEgYAnDYcAhgoDjTWAgZSfi48EjJiHihmAlxqeEpGSqqBXpgyYm5WLHJ 2on6+ToqOxtVmupgi8u5+ZC7GksbLHyU6+tb+Xp8PLE87PyMU5xJyToNbBprjNiKrZrQWO2dKt7d DH2ObiGtSl19m5h9Pe7+/r1NHxAurp/e75+xrp5Ac/bIFbS2K563S7zynUJo8J/EiQf1PSzYQRnD /oSoNu7riBABQYoknVmk52kkg4bwDLb7CLObAnDyStqUGNAhOZUzl4F6yc6jTIz4bhr9lzNlTVdL BQ4MKZKjxXk8j1qdJO9i1IgRuDHtGVMnRqpFdVa9itZP0mRNkbUdurWiS46/3qa9qzYrVLjMVP5k yxKutp128RrOs5Yw15aLGccFq/CxTKWND1vWk3isU80UuAUu53brQppSL5sGlFl00MYna55s6Vaq 16mna/NJLXqz6nC6FnkNHbsi5LO2ixtfWfi48uUriDN/Dp2D8+jUq3f9bD27doDTt3v/Dj68+PHk y5s/jz69+vXs27t/Dz++/Pn069u/jz+//v38/vv7/w9ggAIOSGCBBh6IYIIKLshgg3e1BmGEEk5I YYUWXohhhhpuyGGHHn4IYogijtjaLCSeiGKKKq7IYosuvghjjMndVtkQM0ZVxY1duFZjJDoukdyP OO5QmJBCbvDbB0dmwRY2NOi1JA9RGhFkj65IaSUmsLzAjyFZhtGklo5E1CUUUxZRZUZfNlfkmjaA JMKZXMgpHTAgufkDnULoKQGf3GGnC5VzeXmHn3X91VMteDJhaC9cJTlPomZJWk6bW4oJlKSfwWkP oUlAFBuZ2Mnm21yIitWpo6OKauaiPoA6pGqU7hWrk8MBtmmpucLTmasOVUqaUmZxKmataipx/k+u SZb5K1XNAkvYsB/1SqmxQPqKpbTG8hZrssKmKlk7kk36FK99XYoquemqK66jniLB0kuC3WlrvfQ+ lZC1gfaJbQ6NwhBvvvWuW1qxBh/8bLFhOmWXpdwNnHDC3/IbZ78yTLxwPBALfPC9EFN7naJSYDxt rXTpCy7CV34csbkUP9xxtCw3BWidyLJqZ1DudjsaMh//2FZ3/lo8tM8tp8wxyykfjdxGRq/8MgZJ R0wsyru9SwzOrA6LHLhTKzyo1DUKXbTImj4a2Nc0y9zbbzlHHTLMVLN9tMNKEg2w1k9zu3HPKqvt Ko//5j3y1rim7bfVfauaeN1W2q1J43Pr/pvm3Tf7/LZwi4O9d9iRw0bF4C5kfmfjgEOtMl8Sb416 679sPnDm51r+qd7Vapax5JwvDTJYVojeAumj0X36zkojXPXxvO8rNuwGy768nMCT8Db0s7L99bMa N+/773jfIPzPJbe89u3WcDax68x/HjPSC9/OtNxZY0//9V6P/7zn6piKxfRs1h+z8W1vffXQirrY lTzFwa9371vdySanOpvV7oHyCiDsPOa0+/VOeU/wnwoClrhM2Ut/A9wNTEjjpLLoZYP7c97qtNW3 shyrdjCElgmTx4vDNVB7X1IIsxj1vTfpKoPOahf58Jc6SAXMVqtyWdy4B8HYrYZulRqB/gc9BShI 1RB59xgO/LLnRLLUDF5BrAGsYhhC3SWQTx684vLMWEYWuBEpcRxTD9hYxzCOLo9I4iMK5ugPQJZA kMYDQRv9qD44Xi48hLQiIuVHOzbZkUiPHGQlS9LIit0xjoecpA4yCcntgNKQl9yfjE6JylSqcpWs bKUrLWSiV8pylrSspS1vectYOmEpowxeKYW4yK7N6ZeWJGYLKcnCBWKJNslM3Sd7aMxIvjAGZIPb BKPZTEXOTo+bZOY23zg0aAZzmnvUoiOb0EsCClGcV+DJNdLZtGySEZjOxBoNg9BLGaqzcOxTZjRG Ak90XYyK1MMmKfeEtlIocYrmM6cw/p24lyYe0AQ68mHNzjfRCNZTVt+kXRYvKlGUKGZ7Cq3MDs95 M8SdyhazQRsXQ6XDaKlwMZB7mrS81SR9arKf0HLpr2a6UIrV1AM4HelfigIRjJKlUzMtZPwOmlKg CAuoDXmnznhXQQfmplwPfWJXRUiwrBY0lLtLFQjnRVCjDVVNFHxgWE0nQLcuUYEn3em1TFY6KQ7v b1e1Vu7cl9auJlKKS/NYPV9jzWy6lYcA5GA8n+pPWDQ2f/YjbPtUNyMw2nWc7yOZWevqQq9aj5uk tazzRtvTtxjJp5d1nGtft1FwCjSFrOMl/pTKs8gy5pIBPSZEw2Zb2UaRKQmErGPp/orEzrJus74F 7Lwwt1zRstOeWpKdOauG2xHqFqaV7C1svSi+jYrVpsZlFmrJ69RqNbC4fjqSVTU4syHy7TpITO8M mbrc+ZKzWdbVn1eNe99x3rS+2e1c62ybtPNuF7nbIl50bxg0bL03t/EdMEADK1ii2i56iMMrX8Pb F8X80aDSnOtLy6u7hvbGfejN8D493GDK0dSbjyXrEYdLW7b+sLSzre6Dw3XbvlJYgRweMTr3CuMh RRjAoL1xhf9r3yS39rXUba5fS1XkKJUIygHOsW4TXLL+LvjJFCWxhkF84iUvubBCfqtwicxgFxoW wDyFIocBA184v9jFdZrsfstK/k7ljvnDJ/AuA0WM5h2O8cpIBisC63tcDuLQwYvVcL8mzJlxoHVc lUusjjfN1eeaVqwlRLBGmTvPAhJxhEge14nlcsJk0XZXfK41ZWM862q+mcsZ1RhVad3RKPdxizeF NXZPYUKsQhrCKTD0N91hRL2C9IsJ9aF2ez1jed76tAwdqwSBc2VNB2fQsdUzbD/K0mUh+nDUhrND UZ21g1y7wdJoN7zL3GwzW1l9zo70d3W5h4Zx8pei6ze5qZyngQvD4AkPdpeNDHEirJXOT1J4MBi+ SW3b2JL5FhS/MN3we0sC48+c+LDl+EiSl3tdu9ZmlUOh8n+uCZc0r7nNb47z/pxDSBY677nPfw70 oKuI5/p+eMv5uaMfi5zjOSo64SSJTIebu+I05jXD11rwS08XoVInRSl7a/KVw7HqiTT4hVu8dPpq PJxdZ9LXne5qYXcws6x9FdZTLuGwl7zt/Xu7zM/eTjyBfJlb93ePe6X3v/M98ChvWs5WeipgaZxT jsaUrvec3h1bXtzdG/O7F0grzYmeiSXkqKfPOPoqtvzzU/cl5Gf8rXRD/tXJrn2uQav30uVQ3cT2 9r41z1/5GrX0vLp77/VrXuGjrIsOL+qthl9c6C4+b7yX4Xydb3sft/mIrTbsWY+X+7A8eqssT/uL s3fGrKbPXHefGgbfD2a1/hae80JRtfjXLPCo40ty6o82qT3jX1GkXBhGbsElZ+J3WGRna1G2WA3I f4JXYPuUPn91PwSIKu3HYkNmeBeYJQE1gBnIanmWPz3SZF+ldAVoUsnlYGhnYY/jKyBHgfjFgi5j fMKlWbRnYM1nODt4X/mHTwSWYuFGZaplgaaHcJj3ZSoIgq23dud3f+iHZQcXfS5GfJFihfZmgs0U PmdDHFN4cOW0RkEIYmL2X2uDVCd4dJA1aUu4gdPnhCJYPYj1cVM3WjEVa4oGe3T4YxFIKjPnhWmI cmdYWXrEh2LIZoxjWnoWfriWiOCngICoTDHoWltmTTWIV1C4as5UgkV2/jboxXqMOFgDFX8i6Inb x0OVyBqC5nnzx4aqCE5yCDo25mtpxXx99IicyDMEGG2aCIsSyIP+RBCUiIRc4orzlmGFGGlNVmpo qIeFtF6Qhkffpmw45mM6pl/cNEBVqIG8eIsyJn8zKHDdKHZQJ2WAVoqERmell13152+LeICu1l5a R4WUQYoUt4C4aIzpko0BOIwytoLbaI+DBk94hmYC6DnFqE4V5Gjfx4uTp4SX2GqkJI8tdlTd531+ qIUHWTBzo0aGaI+PF5H9hzsY6WljMjxACGQdCWo46Dhu0wri0lT3GGdT9kI6JY2yGHm3B31FmIZ1 eIcceH0aKXegp3yw/sZsG+iD4YRCOBiU6hWTQghe5CcwEjWDSQiK07RoJ6eVnTeGsseMKHiOUvmT UGmAp9dtUXk1rZeUSpmP3IeHqedkTMhrTFdocFeS2GiXPBY6X/mDRIdS5IhvaEKSkPh0w1SVXDcK 0dh4EWcjrBiQ1JSXjNmG+gdzFgeYdNmYH1d3ducFCiZxkdl0f/lBeOdxaieTLrcFnpmZiSl0rema rwmbsEl0sUmbtWmbt8lKDqKbu8mbvembvwmcwSmcw0mcxWmcx4mcyamcy8mczemczwmd0Smd00md 1Wmd14md2amd28md3emd3wme4Sme40me5Wme54me6ame68me7eme/u8Jn/Epn/NJn/W5nTs3d6ko joHYKiR2jWJ0eb2Jn7skOL0ImXlUFe4UjmhJXNcYIdgZoJSkn5CxTsUUOW5iDiplk6anbqYCmvUR oXsXi5lGfRZqSidKX084NiU1Z6f2nCHKdiMad8SIoDeioCTIkLoxo/hSY9QZjCyqfZkmVfzjQLIm oxEEVJa3fasyfFzJXfJmlJJHoqGCWQWKeD06nc3wQyhEj4+xpUbagisGbmIUFxfhES55hwS5W32Y pj01pfgVkeW3o0eqoy9qhmdZbBz5WSdZe1W1aECakif5ECB0liKkUqYpbsoSqHN6CxoxoVfqpFkK i+w4UVI1FJSq/qigsabWB38jxZX9h4CQyqOcdpEpWKdyyqjc9YnMSYlHhaSy4aKUWn6DYaCd56ok akDVNF5uJqanWmlcoyrPJ6jAB52OaqtihomyeqtYemr4sKyvuhpSKlnQuqNnmIe68Z/Q2lIJdZ3G OpI7qCsjaK24KjgjykzPCqAGNGB/Qq1VGpTZ5qK+amp1CqMK4q0pGRpcCqu82q7MilYRFRnG5qei kpWx+KxnGq3N2qKRGq/6CCXWea8rKYPVyHn8uhmDATqHerEBS6eaWhf96rCl6njEiqoZha/+aqcp SKSouq0OGKro+qZA+a2Req9d+rEbq6LoY1WvAbNzmqMx25w+/iGyVHqFZxp3RtuxAyuWKwuPAIto NzuqEquxiEoZ+kmwv2qyKUtcFZsTfDGwLfq1j8qhgxqnckqo9senp+q1g/qtUxSqvRqyKzqkZbuq yamlyPewTXuuILmhDwphoceyZYKwU9s2w2pR6Vq25rqfPMumdWu34UiwMlqRCRu4SzmyOgWAlBuv HRqlDeuxFQulwLqrR7qhTwo2+2mfqau6q8u6reu6rwu7sSu7s0u7tWu7t4u7uau7u8u7EIqbvwu8 wSu8A9oPw2u8x4u8tYkTEPhypPehAPmF0nST0et7oRiQZXmX46iGOLq8dTa9e9qXsUV3RHiYDkmY iCqQpipp/uorXuy7vc7Il7Rgo0FkgPUaSeUTYp9XUlAVRnryn2tpU9gbwO47wEPJXs8LJszrS2pL T244gnA7mXpZvt+lmmpJwNbbaadJZsWrwItpv2fWhCz5uRGswSQsWNo4wev7byncviusvaIQYes4 Rry3VYCLhdIGXCdkjuRyg6ilRT74etcTttv1li/zevU2q24Lr0e8QknsLHI5cnJLE9lnQzgchUKL hXgbt7WoQ6eIl9KXvluEfacbLB3mVCicKILYezqpZA41xsE3w0WppOuHDvM6fgzTVv/KwEWai2ur tPSnoh55jKYIudLWiLSyi+SFxqF7bIi4bds7iu0byUdY/scEvIvPmI/q+lvE8oH0dsjuRaQVXKWj fFmLnINH51nrO1l2DGis7MXoyMG+iI71C67SRXm1RYtpxMI0m2IA/MqZbIpik27BNslh+WdZvGyJ Bo0IDAY88o2DKEzC44KmwAht+ctD2MuBs4wpvI/QDMeg/MMdKJQqDIcXXMzI7I10JMvYvDPSbMRX BYWh9cqiTKE6y83jvI19WGemrMwzaYwC/MsAfc6PWZnrTIbsV8txg1HxTJNlRc/VmsvErJIN7bgs R8eECFeZl9EJudEY3YgvXNBfzM774s5CpTMHLM8c89D02ou+XDzU6Lnp3I74/L7lrNEfPY2PbMKF 4Myd/ujNPg2+QiUSx3zNB73StKrFifXSysiKm9jP5KyBAt3Rg4zTMed2Z+x+VKTS61bII4yQeqrT P4qh8RuX3dzGYKliSj3VxvyHZt3CNk29PqKyHjteIom2e+zHxwqEaatVgNyPRCnRT016I/kVGxvY Oo3OUP3P5rzWIB3FDRqFS/V8EMl8gOePwHY1Wb17yiOMgK3Wgo24f2hhYVzOTRzVSHzapr3YTi2/ OCrD6/CMCz1UO6uNfio+SlWg2QrLN/1qAfuUjLOZHB1DHm1BVF3cbP20ji3X8zMIVi29MMzMX2DV Hyzd0Y2ZrNm9gmLdgikGzn1XdJS84S3e411zy0ve/ueN3umNSv6h3u3t3jjH3tF9g8r9vRtH381r lear33y3ZvxhdoI8vtMc11XJ2ih6z/BrwMn81guehQOOHv/tWG1Suvv73IfoJeHM3wpe0wxewEiI 0v0B4SH8wHQqPWNtVw+dwbsM0AnuwgS9HiHuwCj8UfzrwKhI2hne4hY8lJ7t3yt6w0HFx9fMw4bo wzP+zjmskUP8ZkWs0CTlxpo7x03MxEvs5JitH/hLlk01pEX9kkS+h5Ftbm8s5sO6rhFOUPmnxjY5 5k8MXWas5nL8zQX+Hv19QG01lmDVqZPp5PcMZsWMyPNNjV29Xwg21YSeiIb+yCvtHnRuxsO1fv6X /tzei2zpSz9+VmB0t8HBetqavtic3spzvekOfh6MnkEHPcvZPK31SHFuDcz3bYRhTtNSjdNZ7s+f vsvtQb7VnA/WnNWnrOhHu8yHXdYyU9GTfeOdLtyIndNETeuZnh+5LttkdlJbfevYFuyf3duVXrCK zI/GzeygLevK3uy7/ewCHu2njosMLaoWjso0bY6Hu+4WLc7QHO7fnu21HtBkPefzfj4fDtPULoVQ 8qXCzurxY9lXmez2Puz4PtDjTsn4geZD7e/TrssoaLVJDWVLrYtNzZP3rti2nvBtHev6vujinFeg qI5+vqAt7tITDUH8nOktP+vgztgz7/HkDvHQ/vS2/6fXcMXXPc3y816PvQ7WKEa3Qh/qxK3wrF7v /r4fs321Fw1tD7jZCXiLKO7uXC6MdwrkGe/z8JraYL/aYv/pcl7yyUTb3W7bXITbF9+3ZF3wQk62 HX9dSA/MZK/1Sq+miU3KPd7c2y2a2A3igL+azUz4BBrff9/dh39k8f3ejw/5suT4kU/5lR8j/2H5 GNK7wwA8gA6ZUO3dm//dEedd8+35oi8InS/In7/goY/6qUb6H2o3rv/6pRn7GbdNtF/7n9nOxMfJ wb2PFH7ymt1Xur/7iFl8LumUSuP7SjwoW97nU+zqxy8HiH7Mv7/JYeayuWpEeM741G/7WPrS/kJu yIPtyeef99MP/m/Q9CQkhidPsa2Iy+i+/pDQ/sufgNS865uH/9JOyKtPAPExdbn9YZSTVntx1pt3 /8EQAoCHVE4jRdZgbdHSldNaTloYVm97eW8i4ZBYNB6RSeWSKdzhgrzDLteLQqWz7LOKjU1pV222 WTaf0Wn1mm1+sq7AL5juOPnkc/K+PuZa2wIFBwkLDQ853vr6qILyGu5YxibhKi29/MQeETk7PT9B QzcU+SZ1ukqhqlAXKflIAF83RWlrbW9xz0hdKYFYd2ccT62iGoljg8Fmc5mbnZ+hGXYb4WBLfsUw W4Ot9aphTYW7l6PLzc/RB6fj2MMugalj/sC9v8mG3VPT9ff5+0GA/XUAGJBgQYP7Bh60kFBhQ4cP PzGEaCLbRIsXMbKxtpFjR48fQYYUOZJkyZIZUaZU2cRkS5cvYcZsuZJmTVoycebUuZNnT58/gQYV OpRo0Xk2lRhVupRpU6dPoUaVihPpEon5MqzjUTEQOW1MrtrhykvX2EvSzH6lN2KsRK1sqyZNu3ZU 2zBz03g9K6FbhD//7Kr1cJQuWop7Cy96C0nYYbGIycZ1gheyhreEDendJpYjRU2U4T6ubLlzYtO9 AovGuhq1Y8aQw0peCJr15K2d3FJexfkHbdUTYqv+yxf0YuKuj//2+1n2kbDBs97FXRxv/rzewwfT ht5bbcLcyJeDD/2aAvbmQwD2vVdx41f10sVb58Zuz/tt9lebV6ZfNOZM9Leapz3B3AMwvGS8KI07 BC1RcA7/bOhrQQjvwmy788STB5xVJPyGMA7vgE85XnzpUIujOpslRWQWFCzEFuFyMMQOS5xxwNpO Oa0SGStUxEZUHMSCxwB9vObG/yz0DcPkGDtGmSergTIcJxn8jUZipJxPRAl3C3Cz8ergr0kWuWys wS4j20s+z8g0c0zF3ERrE1bkbBM2JZcEE0YGc1QTmWFuA/PKVtCUxc4oSZMnFe/6rDKcBA0EjsUv r8OEzkrfiRNTRjRVNNM787SNPEhb/ryMOUcN2+/MUos59VEdebPUVT1f/XTV7Pyr7b5DR+V0tFLm TM3WRfEMtVe6fNDD1AkHekEFX5VtldXNgqMQVCahHXYLE2c7ktK1go2P11S1Nc6eTtM0FgNShsNu 2QebxSdZslR0lZznKESX0Wj5zdRbbEmszsBLwR13z1rTDRjOg9VN9Nhtp+3v1FyZhZhYQu2dVFdV yL0YYITP7Y5bzzj+VeSFRww3ZYMLRpnAhi9gd+KMHWt3Xzz0ndRdjRP29DqPXm4ZWJ3ZY0i/fWU1 2TqCT65P2GzpLRbmX22OmDybBXZBEqbrtVrMJjkD2mTElun64RFdQbrcp0P+VmGn/sV1eeypDzwW RWn9hBpBin+uJ+9dS24N0eW0mzXuoTFGXNI91YbaXIQfV/lhyeemO8PBy8QS80fujrC8QgOt0hhV naYyXXwZ7rhtbTMJHB5fYFVc8MlZbnp21SE33PLZ2BpHnJE7OmvF0OumNsg/fMc4+fxAQrvx1fUe aeyPYn8btSL79RL3YLFPmnS3d79cFXz2KzrJ8bfVLVIPC8Mv4Hllpv7s50GuX0Bv46WY/nqLzp54 2wG1PaLhLXwxk5obLpSRBNpigXFpYMMeKJcDKnCCDKygAy8IwQxKkG4RDIUHawLCPE2FhCU04QlR mEIVDqWA61rhC2EYQxnOkIZA/mlhQGqYQx3ukIckvKE/RBgCzWxMVNkRH0uyhjYlgi9oPWtiejb4 w0MEETDgu8rIGPZAc4mQb/TTxuO+yDZvgFF4upOiOaj4gSE6sXx8+x8Ri8e4KAavesYho9xSh8fT cC2KZyREGo34OUFmyWeV+xitEsEzLy4Rjgn7DiOz2Ec/CgKQAqHO4l4GvAxecYONYmIj74i7Ju7x ktWbZC1eJzQiwctvZbvft8T0Lzb+B14DG9C/Ukm2JFpLS1eLR5DOBCBghumX+JPkKTViljVNyUQz aiPIhse8SEZtcUMaEpGMdE3lLPM11xymFLwpS3AeKUJdzOaHekRAZEYDin47/pfmsBm6Mm1ujQeb lyHn+b1LeU5KuQQccSi3joDaZaBsktvX1pmLdn4PmuIokMHcGMe9mVFOsiJgn7jpGoQWMnckW5tH HUdQ73ksodBYqPYsNlKS3qtYvKTmx1yZuFn27g37q50u9QhAdOEUbqYsqSfiNYVnQS96NCNpt8ZI UXtqzGzilJQsbZpTnt5uqVL9J1Ub+VNOBLV19Wuo15QatkJuVKf+Y2ZEVZfRqvaUdlaVHdOmyket PoOryWKpTEdK1k0VKI+qXJo6V/m5VgF2ZW61Hlyviti35myuzriZO+QqC7D2da/jo+MnD2vLTBrN cFFlqwAN+1eladZ4jW3G/s22xljJZk+vg/vZZYkoWpWeTn1yxORHC/vZtWK1rKU1rUIFxiG+XlWe mkIrISU6S9my7p5wPFoScdtW3fp1p4lV7W9vwVmHuo5gnWuT6WRmQMrGlbTE5G7vkOtZ3pJ3tP6L XO0qid0icBZZx/vM8tLWHs2ExHnQ1VtD+WtFseGzvNyQ7vX6p1IsZtadWZVvV2pLIFxmw33GIF/8 BjxNmDKVw+kcZWDrWGBUWTdw1C1xbwkZ3wcXUYMKUbFFXrziKk4txpQ8pkpqLGNLdvDGW+0xSnKs 47r0kMhFNvKRYyLkDyKZyU12MpOVDIogV2CNU9aiGMtyW1FC8pFn43Ja/qYcZd75WMDdmktrtWzb ySQ4uWU84nrdplhHXlfMiAjzbf1JOKe+caaI/BtgoOpfGIUSzqyRsx3DWuc23PmQkAwjV3J15THP +KVt9jJmDV3KS4uS0Yr2s42pLOg0aZLSaU6k7T69ZUzPOdRvrq+n7azM9SFYmL8jqoG96lvLjpeY tbble1xqt1n70tfmfGM5u8fLb9ISUs38Mawdllt/OZtP+h3ghnim6x0letuD0uaKtCnsEPey18YO 54J/ZKZlpzuY54N2IU6K63dmqTTehRN4fVNMXqNUuLrlZ3H1pFYNHZSg4xKowQtO8H2/G4nFG9Rq /xdAqh53ZeQj8EUb/oXRYWuU2ym+qcdDm/HqRhy+z2b4pK2EzpUStlZoruhsYyvir/Y5bBQWNeV2 e+gCvzennT75h00xVK+aLa8df7mbxw3zpi4YoKTecMjdK0acm1jbP19DXeWlSKJN1qcopbfRGTzb GqFuqrSaOopZrWCpl9zqEE6OXbWO12G5HLkQJ7BZZ84+wZ5YvTqP+uEKHfajtj0vWcPZtfLLdaDX HZvGNHVM7X7iDHkS8H5Xe+VHLvhKE77w1Uwt4im/Orq7McCYXW5RHY34mP9d2paPLminS3POs9ig DE08cavtp+Ci3PQyX7opn/v4jy+2vZdvPZ1nj8C9pxd2u8I3PbPt/mCXn56ozb0beo998+Frnuex Z2/Vk99w4b829L3kH7Zh2bz5aZ/EHsrw62C7erV3b230D6n9oUft8Gd5/K9t3yuHBn4gTf04jf2I r2Tsq+lsrv/yi82KygGrD/8AbPH2TwR8zqRMDt4yEMY2sAINqSousAxCsB9G8OdKkBlOUC5OKQXf 7clc8AVhMIY8UARjsAZt8AaZYgZpEAefQgd9cCJ4ECp+cAgbIr7qSfkybdWIcAlPy+SOkAbjLPqY cArZyQmlsPOS0MGocAulzAqVUAXxrAO5cAyTwAivEA3oSwvJcA3/KDWCja1qatZqpGJI5/xujQ3x cMkOI9zgcAGb/ib+TmTscge/WDAPDfHLpg4beuXhOseytuT6iusJD3ES1UCglM5N7I8RAcvd+mmA vpASQdEq1o649KLfxmnQWC4PdiYUWbENMS+p0k+otCalkE7XmusOWzEXkfDA5GO/so4W8WgVK60Q dTEPu896cs1i4A4VXSQWhLEYoREKjw8Wk/Dwjqr8VNGoonEbwZAXI2WIIkGo7kruoOUZufEc52sU DYV1Bo6hKG9NILEBPxEd6dHSZIeZ2HGsJI4TO5FzSmQe6zEgVe0eowkXy67x6osQf+dF1FAgHbL4 dMniqmzYAq1OIlAAHzIjowNmiFEjdbAj9dAjRbLRzgMkRzL8/oKwB09yJeskJY2CJWEyJmVyJmmy Jm3yJnEyJ3VyJ3myJ33yJ4EyKIVyKImyKI3yKJEyKZVyKZmyKZ3yKaEyKqVyKqmyKq3yKrEyK7Vy K7lyhNTJJLtyJz0JLMMyJ8dSDMsyLdVyLdmyLd3yLeEyLuVyLumyLu3yLvEyL/VyL/myL/3yLwEz MAVzMAmzMA3zMBEzMRVzMRmzMR3zMSEzMiVzMimzMi3zMjEzMzVzMzmzMz3zM0EzNEUTKSkukdBy iqDonABxNFHQDE/TzMwn0jTp4LRkNVnTgqzwNZ/KvmBLfmxv4PJFN29TiMgSCeytDhuR5MwJ22wu A0sTN2tu/tl2EDdnM8NW4jn1IUlMZ6IGS+/8T39UDDvLcIFITySS6cfKM7isU3poojSdrTuzD4vk 80c4yrUCJV8cBZ2q05kqKx2F8+iGLFYaDA3TiOko7kbIycJCKDUR9DjzM/6ssza1E2wQcHQm9En+ gjkB1AjE0zjJM342FIF6DN2OqzoZr0NxCP4qpB9dC9io5D3hic88Cj8BVEOjxEYlzz/bsILSM0Sn kwgmbM8Yr9qasz3hr0X58+u2S3QkTkaXD1XAEz41TjVtE0h3I0HboR0RJd0GKzazSUD5LkjTx1Du y/36syU1BK2YLkdH7CIiikZrk7sGMEs9Tmoyx9a8E0qS/pM5CdBDF7JB6bTxHIpPvYtb+rQ+iWc9 +RNHFfUYDLWdehSkCMdH3dQSS0xMnRRKQQdFvQ7XjAlLI/E2opSD9Kn5oI/fFEQ/VxQ5I/VS/7GN 0mlVr0RWL5Q3WpVCAepMMeJNS4/PkjNE9ZNN9SxN1w1N/PFLKdUCN9VUmXRJH3VJm3VAHZE3kfQe XHW7glW82jRTuXVahRQID+79nnVINRVat7XmxgRZj64R2bVIctM+f5VIPUU2Y/Q+RyddI/RaOQpO 74dH61VYma/ezNNIRYqmAvVWE7VJmy7U5pRF29V8xtNc4zVe7bVM4/NicTVHeSRQMZY7OVVeMxZY 6WhU/q/TUm11H0FUX+2TppCKX0HkRt01YuG1S1RVbOgVVm/pXpO1UGd1X0uPZ8kuU9XUUpezZGVN Svm0WulpZuFzYUO2X9EnYYM1St+1TrcUREYWeXAW2BB1SPv138rVWzOHWBl2Zc0WYLkzZHdVRVlV SVvpON9TZXWVVU21Z5m2XMNzWbEVdBCVYusTYWG2UOC0W5dTfQ6VXDu1QiEWx3pVfu60Tj8VU922 JS/LNyVUbzmGPWXWas30bOvObyUPcDG0SQf3XCsWbcWWZCU1aouWcQ8XVJfWES2ysrKV/HgzDsf1 bnt1c5VTdsNURJQ2dhP3bH0HTzuWc9PWc3NVbfsm/k1ldE0fs3bj875eVGtL9VX3LDhpVXqn93pR lW/BtGudFyHx9j/XknuDk3JBF1BvVzo193KZ1VfF1VHj9qnCNzqLV1C/FTLRV0HTlU1htDmT1Hb3 M1SFVXLjFHWVFzsrN1ahdzghOIIleIIpuIIt+IIxOIM1eIM5uIM9+INBOIRFeIRJuDNd8oRRuIeW MoVZuIVl0H7dUVUbbFaNV95s2IVhgkC1VY2wjEMniNCUlSQxLc9SzblkrYivtYZpeIZlOIaxFoff V/xardQWTloDLIEQbchS9cyOOBaRI7y+eAAFZeNQ6QolcZDGM4j5b4p52NW81f1ydsdSTosbVIi9 /g9HRi7PwHiOKZDMslD27FGNqRgs8k0Mu8x0AbAu2NiFUo/m9tjUPlCaFrkh1cGMz5CRu7GN1xiN M9nshAWLNQ2TBzKSaYv3+riUJxmQNRCNuWgDPSjMGJRMvfShLgzLMtENsaaWCpUeHlliZmY+5Tee mOWXwTOYyc2IUZOVJenFXlk3VZQhyXday0ghJ88BtSIBz+1rH86LY8TxjllIvPmb2K3d4vCbkUTO Nou6ZLN0apl1Lfl5ExnExjdPVXnThtWU5RSeajUeHTQ2ynm07PCzmDUex4mMo/b2mIfoaEuhMUzu BC4xlgnZqIFnBxH9qjFyoTmBpRlnAXKUDdY9/tmGoBFa9Jp2I9OqYO/N+JCRWNGZ5GqRduMkZRXu aWPaoE3DFJntQbkWVfvRkrnHe3v3dPV0juz0cG96ZqwmVehOVx+6F3fukuX0k5yaar4xj6uaqqnx Hk8thiOPH3NvHTMN487rz4CxqwaZYQf2qOkQ5kil6zy3qa9apfd3VLjqpXv5du6aTvK6pt16EVXO PLDRGbUR9qoG77CaOOk4iu/arPPxVk55eAdPlbjvT5XkFmHvjpFRsaaam2ealDVK6Cx7n4zKHNeq sNm6fKrYjj/aJFC5rU/bvNyaVylrs18Ptcesrjt7twqNtls7tymZlboKsFNRsJO6tJHasOH4/uJq qL9WrmbADnHXWdom+35DA7cFWozD6LDtGru3G/hwWRlnJbBh+7UVx7Qbe54ZZYZ6e7zPu6z7bGhH mbfzr4dfyrrxmrt9G9U0O6612tUYEpqFuxnZ+7/+8LjJG1OaObHT+o8BjqxbrqSx73KOcbr7u/08 +aUle7+zer3v27tj5POucbgH3CDL28DbW6mJWsHzdc7wrV5/tW5FGlaKWbpPj39SuhOfzq14e68x XL87JQ3FiqsdXB4vMgrNSqTf0eLcW5FJw3I9uiA1OrnvzaJZXFztWb4r+nJ1L0FznJ0vG8G+HKfL DqdBqH9d/GD5+VXTzl+AWqeHukIF2YWk/vPD3Eeed23LSRx+q2vCqY+i7+ShSRHCAx2A+RueN5xJ ae99HpX8KhrA1TGenth2t7ajU1tLOZmCzqE4WzPRkedrKqxL8xwe43k+/td3f7uvZVGUgcx80YPV gcjVKxHWo02Td3HVM13WESLCykHTPRqfJQiKgT3YcxCSm7DYEbsshD3Zld2G9i54nB3cijfaCZVL 43TZT2KTid2k7TnRd7jXz3q8gHzbp6eLCZRKzb3aoX3az/3Zrc3aF1yKL32rxd3S9deyfd2wVvsI 9Ri/O2axs/uxU9TIT73SZz2Qaj3bVT2159dmf/hptsM37fvQR0nDwxjVSfCdB97i713b/qUxlZ0j lHlVyRMeiY8o4ku+cBr5IDH9j1tZFOMc3hG+1eN9MUDZ4+O9cjgp5Tm7y1MCyFu+k+X44Gd3PUpd edqZufsdl39Zl+dwuLwdbMvtn6NbnIE36jPPQ5M5DGUe6OW94/GV1jpd5Z4JHJXK77CZy8nNyrV7 D8NZnMbZ6g0jAb+Z0STwr6YeTwOa7OHZ+khUewFe51/eZ8Y2qL+aSKH6/54IyQkaG5xVWt7wTeyu oSN/TrcO97we1dJn0Smae6j8mjHaSKR8owWV0jWeyYWt8vHKxk2fl1G6dOTaoNU+wq87fPW6Ybv7 72kdFYEkfgX2+SLRp7+rzfcZbEc9/ugvX8CH8bs1fsYhUo8kh5P0mYs3nMd1vNABPdaZuKuT/E+a 1mSLXPtde8S7HuaRn7SBG/eNLfEl/vkfmK6l/98l+fbD/MdxPam6IMBBT+STscTXur39veAJID6m LidAC29NCVW1UXN3N/V82HhkzGmcabeOIgyiFzuTiBuxuVl6dw/nq8mKRlCGFuwIm86fiuZrTZ/L WgX7IUZFoWORCx5HddPseVu9psle8xcOLc95njcYHR6++nKrXRyg3xyZoZsUnR4Q3VJdIqMYk2Lb ZKMkDJ5lg2Tj4aeYlqcjJmTh3lqXoE3hjpoVSSpr5yhTIKOjbarrIO4nqOyicOWo/iilYCmQsfFm CB5tV6jsL9JuJXPyJnTPNFF27yOsrzgHNG+uNO7tqa74NjVnN4ReEvEktunTN1scft6ikUz24P3b kOJZiYMJXxV7pykeK4it+KBjSK5WvT3qCEWseBEjR0vvCI678YLKKoXI8pEKRlEby0vTlM10pmkk yVp3uKnpuVOVzJ4WO8arss/TuXYNh97MKCOpUqJRm4EE9zGnQXkubqk8hTCXhaGrZP68txDKOZx3 vKgl2Umgrkw2mwqJ4bSowHVgx627G3Ze3pB/Y+yVGogw35CI2R2mi/Uxt7plH7oS2tKyTnc+o0UG 2jns0sxj4ELWWA1gtDVyYRH2/juXtOijU8lKTn021stqG8syFvxwL1e9palBtZ27Llumcl0nRj5h 62KeCPMWBhZ9ODvs2ku33Q6vu/fwf8QfKp56LE/a6VGOX8py9Vzcwq3D9w6ePP6r+bHe30++v38B jlaTgKblVM91uh0BYIHaMdggehBKSJV4D05Y3YUGusVHgjMQOFhgIYo4IoklmngiiimquCKLLbr4 IowxygjbUx9mqOCNAz7GBY0emjMjkEEKOSSRRRp5JJIqLvgcXNQ52SSUy0Vp15RJWmnlab8ZVyMi /LmUH4CyOVjTWx+WuSSOXGaV5mkgAiYlnFTGmdycTNZ5JZ5EBlSfm2KySdyX/gNWFphoaI53X4gY ZjdReylZk2Y6awJnY476BEohMP9R+meWaBAqUUEHXopKfSMWuqekp3aK2I+QPpoqhWcqKmtVlXZ5 KnMEWcjpL92lKJWamobqmYeqBqvhrcBGCGtJzC7aqH7o7VpgrpZyt2l52EL7T4nLOrudQ8et9J2N iL7K67bGPoupt+w2Z6uOqKobb3jThnniuusppOV5rL1RbWE8CkbTNWnBx6dZ6aK1DMKUFNyhp/00 XA7Di4UL78XzDmuftqBmi2K0UDqnRal9UKkoTB4n06OdjpH2aUMKN8GyFBC/SVnN/LbcVY9B0Gxb tB+ZhyDDyKnHGXsS6fzv/kL8DiZfvsl+/Cum020GWm1YU4GZu8t2hRdYrCp2tdYq/faja/P9FfbA Rsc861dWofzs0Dmf/aZ0gFV0Mq4375vcyGmUDJu9HRtnYtVNs+2VRfsC6nHitxGbNOWFwtw1znLH wijcks/G21TmKQydE9OtFTHSW68XuXJcm5410yZrjCwiLkKLWkuSMmPI0s1OqplHs5cNcG+aFx+8 YWM32pjvUlv8nuL+Nh595OHA7nnCIsX0eICRLjyuHLsLCnniu1kfFHjEA4+8tMqTz/z77gctb+oa 7P61tSn3TZPE4CsrPLo4drvtqQ93AFwbsGTTF+UEjnfba5/x1oeUc7HP/nsVFBf3JKACfpgvYAzx x/IIKMLOiY9cArIgB883i4Gg4lsKHJhfSKYWAyYvgr9j37smGELj7WoYx+igTpwCwvgFcYR9K1lA aCcs8lHMfytsV7HMwMLj1epzWZNXqzhnueOoDSpd5CIMzRQoH+rvbdYbYg31h0Ypqi9o0zJUAo1Y vjK28Ftz5CED01W0Y+kQbkBEYego2Mcq8vEL9BDLIIuYwvwRUY3EGAnTljQqME0ybotjpOP2NBNv TPJzMUSkwCJBJjBexYttu2Eic+hJMermJLkKDvjW+K49qtA0kATkEouyutd9xpJWQxkterdDVNoh k3G5Hu3gl8ZUUtGK/qhcJRzBtrrsPXN4bEMkFY2Jv+wZTGpRw06Z+LYl+TRGnKVIFIZeKEHSodM9 z0vm2Dj5R2HKzWbzVNvHrjg87O1zeq7rJIJ2GdDrQe+BvHoj/djoTrxgrzXYDNw7/0fIpGiTTzJs ngTNYhRS6vOZxmzmR99YN4TJkEl7A9z+JiJON71GKAfT1t1OaDh46WqmtkIoTallU1VKx2uRdM/R IOi2v6nmOjRK3zbxg9OcKpGphXQqVKEY1QxOFaPHWmoAqwoZrMpUq16d2/y0ytV62bSN4MwTWtOq 1rWyta0xiuZXJRpVBqHUP269K17zqte9CilUzyPdNP9qssG6LXV8/j3sTh3oTapKdUdl7eSYmhpW dtEqrlkBbP0EW1gEZrYpREPsXXW62F5BNkvkXM29YGUu6qirsrjUIlgtK9vZUrKr9KqpHRnq2U+l 9lCGa2cWfSszaNK2uMaVaWI3llCybsxpQOntcm9rVeG9dpjHvS52OZbc6D41stH9BnQlK97gcje2 Gc0uetMrqtqaVqMCy9xQH+pBlY2xYgSzmriqu7CHCVNLnnKMegMs4EzlUrdxSRBrJ2hOyIYSmH47 y1HtJmFYMtGQFpWwDX4GoskOuMMedq6XLitQsu3zMgJV7FjOKRaxWcWzi4tpjdJ2SsN6TnQevjGO O3otZlnSjOr5/iiKU+ySGn+tmBiE6Pzgy0x9KtO8OX5yeseKSxoWUI4O3ChGTSlKRmqyl6Ocp3WJ C+UxP5mrU5ZjNkooyUayWXvIm5iByJtIQDb5gGS+s2VBTNoBKjKlao5xUaGoZRVCR89PLGSdQRpG PDP6wzg9c5/n+GcR8znMFG0bnMP3ZdjSWX5ObjSo44pQSJcxzVOktC6zzFEuS4NSNBTqkj266FDT Osrb3WL/tqzpxup4Ueok5oxjUzB42nCis641sq9raLj6osfU/DH1YtzdX1fRyL67NLFxqOjzJrvb sl22ePvpZer5kmgOlragV527dXv5is5u8xfB/E1v09upDwpn/l1ZetpBLPiyl2swvHPDzn83DYmI jieWOQ3gejP8q3Td5E976qOGRlzFCaYKtd0cGqCRw6ib1hw9cbjwhpN8quCucKPHatdbr7zkLq+q 0yDubZXvh+befTnO7S24rN7Z5kplec2BnvOhh9g5/gYt0pOudD0Rvek3/a/Fly71qVMdcU6/eo5M hfWtc73r6BWR18Mu9rHPVetkPzva075ys6u97W5/e027Bfe5073uR2e73fOud7hbfe9+/3va5Q74 wRO+6yQqPOIT33S8K77xjp/55R4v+cnT+uKUvzzmx9yzzHO+8wHukOdDL/rinnz0pj89TX2O+tWz Xletfz3sSWMv+9kPHX+qpz3uPW97oee+96ffve+DL/zhE7/4xj8+8pOv/OUzv/nOfz70oy/96VO/ +ta/Pvazr/3tc7/73v8++MMv/vHDqwAAOw== ------=_NextPart_000_0000_01C85294.643E2190 Content-Type: image/gif Content-Transfer-Encoding: base64 Content-Location: http://infolab.stanford.edu/~backrub/red.gif R0lGODdhBQAEAPAAAP8AAAAAACwAAAAABQAEAAACBISPmQUAOw== ------=_NextPart_000_0000_01C85294.643E2190 Content-Type: image/gif Content-Transfer-Encoding: base64 Content-Location: http://infolab.stanford.edu/~backrub/white.gif R0lGODdhAQABAPAAAL6+vgAAACwAAAAAAQABAAACAkQBADs= ------=_NextPart_000_0000_01C85294.643E2190 Content-Type: image/jpeg Content-Transfer-Encoding: base64 Content-Location: http://infolab.stanford.edu/~backrub/sergey.jpg /9j/4AAQSkZJRgABAQEAkACQAAD/2wBDAAkGBwgHBgkIBwgKCgkLDRYPDQwMDRsUFRAWIB0iIiAd Hx8kKDQsJCYxJx8fLT0tMTU3Ojo6Iys/RD84QzQ5Ojf/2wBDAQoKCg0MDRoPDxo3JR8lNzc3Nzc3 Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzf/wAARCAElAMwDASIA AhEBAxEB/8QAHwAAAQUBAQEBAQEAAAAAAAAAAAECAwQFBgcICQoL/8QAtRAAAgEDAwIEAwUFBAQA AAF9AQIDAAQRBRIhMUEGE1FhByJxFDKBkaEII0KxwRVS0fAkM2JyggkKFhcYGRolJicoKSo0NTY3 ODk6Q0RFRkdISUpTVFVWV1hZWmNkZWZnaGlqc3R1dnd4eXqDhIWGh4iJipKTlJWWl5iZmqKjpKWm p6ipqrKztLW2t7i5usLDxMXGx8jJytLT1NXW19jZ2uHi4+Tl5ufo6erx8vP09fb3+Pn6/8QAHwEA AwEBAQEBAQEBAQAAAAAAAAECAwQFBgcICQoL/8QAtREAAgECBAQDBAcFBAQAAQJ3AAECAxEEBSEx BhJBUQdhcRMiMoEIFEKRobHBCSMzUvAVYnLRChYkNOEl8RcYGRomJygpKjU2Nzg5OkNERUZHSElK U1RVVldYWVpjZGVmZ2hpanN0dXZ3eHl6goOEhYaHiImKkpOUlZaXmJmaoqOkpaanqKmqsrO0tba3 uLm6wsPExcbHyMnK0tPU1dbX2Nna4uPk5ebn6Onq8vP09fb3+Pn6/9oADAMBAAIRAxEAPwDpP1op cUlcZqHr2pO+ace9JVXASloooEFN+lP/AJU2gA70UUd6oQ3pWTrupy6aqSRxRsEYbmc9/wC6o9cH /wAerYrM1GGOM/bltftFyOEZm+59M9Pw60pJvYDkLszW0mJLKNVn/wCXdSd0YbH3R24/nWn4ktLe eyW+2rBOx+cDo2e/4Y/Wr8en22tRLdXVs0EzSkzpu+b5RtC57KfvVTa2SBX05ZUjeHdGD5e9mG3J IHReP4j+FZ8rV/MVzG0Q6jq0N1pq3BFvbRGdQx/jzhV3enU/gaqWWo3GlyvtkeN3GHB7rwe/TjvU 2nTT/wBoyQWl5FbRtExLttjVl4zznHf1NLc28+u6mCm1pW/ebm4jbBC/gOg9/epu9CdL2JBNdX0s hv8AU3ibZuRdpIYdsVXAgu3tksvtLXcn7tt7DaGPocc47mtG+1V1F3aa1YRCVpfMyrYKbsMdo6/e OePU9qdp93pmkQi6gjSdpMpDJ5o3/eJJYfwDGOO/tVNX3DQfBouyBorn93qMZOx9u9bhSw27fcdP XrwayvsiXUptotrymQhXj5zgHjj6Vel165JkgubyFU81v3qx52j29sdBWdCbvTNTiuYbmSO1eUuJ nAb5ful8fMCcZ9aThF7aDuQxQmCRJ0JPKknG3axzjn+oqzBeR6dK0qHBIZS27IZf7pXpj/GrmrWA istPuwCs8tkssrf323Bjn/vqqMFhBc6cZTdKJkbLw9cKT6cf+O5rO+pVuiN24v4JtO+a0lgKf6w7 MLkJgH16Guz0pnk0+3M4Il8td+fXAzXmsLand3aWyyGe4U4Rd+Sy+nv8v+PrWzZ61NZXqNd3F6sH dXfdxnjn+Kkpq+rEro77vRUNncwXkCz2kyTRHoy/54+lWK1NblH1oxyaX8KDVEidzij19qXvSetU IT2pKU1HcTQ26briRY1/2u9MB/c4orHuvFOk2xbMzSH/AGFqifHGn7ji3mI9eKnmQuZHTH7xo7Vy x8awHJWyc/V//rVt6fqUN7pMN/uWNZB8+T91ujD86aknoguXCaqS2okvTcT4aOJB5anoh53N9elY mqeLoICwtU8xgcZbgZrnrnxXfXbCIS+UjD+AYznrzVAdVcahDJJd/ZJSDGPLh8heWYrnc3t2rkLu 5nu7pvNVG5+64AOPYVWhvXNz5qAxZx0Pf/JplykshPzO7g8AJ1X8KXK+orDr6zVdzoxG/ayBu/rz VYbrbaAxUv8AxfSnu8qnYVzz/Dzj2q5LezSsrNaxlzjDJ/PFZuPYfKWL+OyubS2Fr9qudTkn2yTX AO6YkH3x94qPpUen6Hefbltr2KaGLz/KeZMHaSPl/Pj86pzkTFUTcmCcK/O38frzV+drS5N1INQu YbwLuTexYSt9euaq19xOLR0y+HofI8gzSRx9B5SqrEZ4LNj5j/n3qhf6VBbaXMLxh9sRwUuD8vnL kDAHThcD/wDXVCw8TahbWxVmS4UbvnnBLKfr6fWs28uLiaOWe6Zn3scv1zt+Xn0GW+n5USlB7IlX W5La3N1Iy6fJep5CxGNRMSUT057CtnThBHoiv5EUt1AWeSJjw+T8rH+8PmHt+IrnLLE0ksUrrEAh JZ+MbQf59Pyqe5vZHVLaYxgQII4ivGFGSP8A0I1kUtDo9R/069hvEZp/LbE0lt/rd2zOV/4Fj/gI 9cZXQrOO71tg11K7RKP3Nxg+YnGCD0OKpeHJbpZ2FhL5csZD/ZHHyyKcc/Xn+R5rqLiFbbWbabyg NiMXbO1Ywcjn8f50rXdxrU1LbT47S5aW0HlrJy6jox+laFZV1JdypLFYy+Y4b52XCkf4VnLrGu2i iGbRVnZf+Wgk6inzJFmz3NJ70vfFJ71qIOnWj2pevHrXMeJ9c+z5tYWxngkfxH/D/CplJJXFYk17 xPFY7o7XDy/3vf2FcJd6hdXs7SzyNu92qOQk3RaU5NQvkHd2rJPm3J1GYGTSgqF9GzSSgBuDweBT FQO3znCjrWiVxJF2yVph8wG3uDU9zezj92p2LjPlqf8ACqkF3F5nl4Iizx+NT3hm+7G3IGVZR/DW 0YW1KsKD9kjWUyWxfpswfxzUDotxOqn5Qxz6CpBCkdstxfN5jls7d/OOn86pSztnO7DdsdqsCaVh HkuGCEnYAcE89arb43Xf5mGBxtqsCJJY43LLFySRyRUp8hZcZA+XkAdKqxN2W/PUhQ5baeevPT/G qUzybz5MvyjuaS9KTksnBA6U22j2ou9s98VNiibfKABnLDvR9qlxjr/P602doo9/XngBqr/aIeyE 9aVgNazvVEcsc8UciyDGW6of9mrM1q0M8SQyLL5gyoifcOvT61iJIk29ujDpVyCZoT5kZ6DrUSgx otNKUgVVxhvlDei1daOW205CuzbLuflAclW2n3/pzVeFxNGzK+ZZG2lWXIC53D/x79Kk02MS6jbR XYAiEy+bub+HdyD7VzvcJWRftQb0QtGGWKNdmwzAODtbPlk/TpwPmwM1astUu/8ASWa98xYkXaLh Adw6bWHpj09z2qvcNp32mT7N+9sfNZ/JGVLE9Nvpyx/DFa3hCC3j1q8itjMYjbqJknTDKwb+R5qk ruxOpoQWkVvrun6hp0ItI7lCZ4lJCrlc8r0HzFa68kA4qt5Ee4hl4PWpxwMVpa2hSRV7HrR0NJSZ pjFyFy3XHNeXeJpSbyPPZc16hwSQe/WvNPFlm8EscjD7n7t/6fn1rKothdDKHzbuTuU8U1j97H3T RGd688MBTJDjJPDenvWaIIW6be/ap3jHl+VHy2Pmf0qG3wHeV/4eBmnIpJl5wqpl2zXTCNxjrO3i VnCSbyejj+E/41qt5SpIlxGdoIG1Wxz6AVi2s/k8Hby27DevQfgM0lxLcXciCPcsacJj9T9a6NNx a9DUvEkMDeTA+Dje5IzkfeOB0HvVWTT0QRF9waTjYV+6M+tRROyyNBbzKm7h5X9vSrkKO8DJLdSS NwgVsj9PyoAjnsxgSIuYI8bmTj739axL9Sj7VRgT2710LQNJYRsT5UAyUGctIRUMlp8imMGSUH5n PY+lIZz6hh8zPVu1xHJudxtPrV1rNHOXaNcjJHpTUs967Ytnfkc9KQFScRMNqAbfrVLy23cDAq/J E1uJDJkHpyKpgnLBjkds0IBzRlDjG0Dj3qRpDhEC4AHX1qEbSwEgwPULUm8Ffp0qgNnw/t+2Ro7B GdvlZuRmvSLWx0+7t/Ml063Eh4dCg615ba4dVHzBvUmvSPCN4buwffjfG2M9zWaSY2aP9j6WOlhb df8AnmKswWsFvI0kESpIRhmTgtUvc0761XIieZj97+pz9aXzG9T+dNAp1PkQXG/y9qPejPX0rNku xbXsPmN8sj/ZZD6P96M/8CBYfXFYFmj2NY/iPTlvbORtv8GHx1I9fqK1+9Vr+7isrWWeX7oHAP8A FQ433A8oeCSG4kil6x9x3qBv3h69K61tMSWI3F0pDMOEzjaO2feqraXZdPK4Puc1zN2YKJzjA5jt oxmQkD8zSXqlJpo02hVYDC98cUsTkamHQZ2Px+FT6iT9tlkeNVVeuPu5/wAiuun8NxPQrW1p8+3B dcZz7U5jJLP9nhDBc4b3rf8ACtsl0/mMMr2X8a3P+ERX7Y0sUhT5wwCDFNyaKUUzAvdM8mDywu1Y os/MBuZvp60/T9NlgRW2D7Qcrt6rGpHf3P8ASuytvDlv5pkmLN/s5rbg06MBY0iCRr0VaTncv2Zx FroQkDRXW6GJT+7b7oPv9faqd5pXklorS48wlu+Pl9BnvXqMunxTwFHUFT/DVR/DtvsVbeNV2nqR mlzsv2Z5xY6DNqN55LZ+zRxZcr13dlH69KuvpjWkXl2dowkY/OrJnH0PpXotppMVmw8te/JPU1PL Z7gQBil7TW41SPD9S07UvOZZLeYlvmxtzVFNKVkgRZFa5lPzL2jX3r3eXTlcMGTtWPqOhQ72ZIE3 Hn7vWl7UPYs8dvLAid4kKuq/xp0NVI4d3yr97Feo3OjJg/uhxXA38P2HUpY1+4ePzqozuZShYZb2 4w8kvG19ox7V0Ogax9juLkbW2sqt+5XPNc9qW63iji8sop4z2NdP4P8AMtXup1HzbVBBGa2grysZ Tdkan/CSrj7tyP8AthTh4nQf89x9Ya0p5vtbq7RpCeh2DrULx4dQDuB56V2ex8zh9u72sVR4ni6b 5PxgNL/wlMI6ynP/AFwatKexM37y2jMMW3lmO5SfaoksJ5huj8vb061Hs2b8yLlnfWmoRtJYXCTo Ouz7y/UHkfjWJ4vhMlpKFJBnh4Yfwupyrfqa4OGOSGeO5sZmt5gflaM421p6l4kvLqya31JFWZR8 ksYwG98f5+leW5KSstzoi9Tv9K1BNS0u21DcqrNF5j46I38Q/Btw/CsgFtcv3lYf6FbHaiH+Jv8A PJ/4DXLeGzqVxaRadZ3T7ZOqdViTOWyP1+rAV27S6fpVosHnxxrGMBd2W65Ofc/1rRvmYtihqjYO M981lOT2FR6lrlrJL+5DSkf3RWa815c8JGIV9T1rlmm5GiehhOdrS3H+0QKt2MqzRPbyAHjdub+H 0pbm28u4eJuTyTSLHsIXdgHnj2//AF13QfukPc6/wrbm3l27cDaP/wBVd1AuRiuR8JQnyt7A8ev+ fSuwg4JzWUjeKJ44wPrVyBKrpn9KuR5BqDZFlY1C4/Cp0TA9TUBDHgEirChs4qhBjnpnNKw69iDT 2HWo8EfeFJ6DWoxl657VTnQZzV1885NU5uOaxlY0Rl3kKsG4rzTxhp+GZ8YFepTgY5rkPFdqJLRv c4z6UoSsyZq5wEskVzpah5MtGe9bvgsmVLrLAEvnrjmudNtGpIT7oHJ/vCtrRcLabgcAtnb+NejQ jeZ51d+5Y60xElVQ5J4Gw0PC8Em2TcG+orLgu47aMI/DE1Y/tCMne+48dTzXp2PKcWaAW5EBbdMI T+VNCXLfMgBB9GqkmpxsDGHYqP4azp2ZpWKFgpPFNRLjFs50KvUHj9aqajKQobaSycgYqVH2b1Q5 Y9yTmomJ+bPJHrXzUI2dz0+ax0el+EdUAYXVyLIE/PFF8zfQ9uPxrbi8KaZAuZRJcv6zPn9OlJ4I u5LnR3ikJItpjEjH+7tBA/Ddj8q3j0NdkUmrhc5TU7WG2x5MSooP8AxWXznHeum1WIMjCucbGP0r lrxSZpB6Gbq+2OW3kblcEH+dQQQmW8TAJLN8iepq3qwV40Lf3jWr4PtllvpLg8+SAF/Gt6T/AHYv tHXaZaLaQ7V5P8TetacB+YDsapg9ai/tGOF9gBkf+6tKxtc6BMZ+oq7EPXrXHTapfBm227KP4TVe TxPc22BPE/02EVooXD2h6KiZ+6M1YSM4rhdO8XN5qh1LKTgg8Y+ldPp2sxXV9LFnCn5k4/hquRIl zubG3OCO9RMpxn+tTl0AJ3DisXUtajiWUKCGjGT703FExbLpAC49aqz8gFuhrk9T8T3Ado7VflHU mqMGq6xdltkEhH97tWLpo3VWx1k45ZcViX0SzxNG/Q1BFNqSqxkjIamLetLKYplKS4yP9r1xWMoO JfMmea6vE9rNLA33kPf8MV3uhaMJNOiX5V2j5lK1zPi6HGsxnA2ttJH0q6uvso4gUD2zXpYROep5 uKutEdFe6fZW4zdyQL6bqamnW9zF5lo8Dr6qc1zb6wZG/exbiPVqdDrPk5Mce3Iwdpxmu/lZxWl2 Leo2i2l35YVRlf4KrHccEHtSy6rHM2Z4d5HGTQ1/bE5+zAVSuik2uhyLc5z0FSNC/JOdtOgxlo3A P901OpELiMt+7z8p6ba+Yvbc7bGhY+In0S2isY7eORVLMSSc5LZ6/jj8K6HSfFOn364lb7LMP4ZD lfwb/HFeebVeZtxx2o8sK+0qw91rphJpCPRtRvrMoxW7gP0cVzXmRyEhJFJ9Aa5uaNQSPMyaoMSC dgJPuaU4c/UuMrHXalayf2YbplIiD/K3v0rS8CyK0lztHHy/jWlpVql74R0sY82G5t1Dr33L8rEf QqaoeG7b7Brt5aKwMRCvGfUf5NOMeVWNba3OquMqGYfe24H1p2l2iW0WWOX5yzf1qYR+cMYzg0y9 3rC2yJnYfrQmVYlm1WPLBAoUcF3+7SHUNPuNsUs8LFvu49fSszRYV3SyagMSshRCRlYf90fj1rC0 Lw5qK3VtbNawhIdRW5fUMZkKAY8sf7PG7Hqfat0rkXOue0twN/lxyJ/eQdKltI4xOphOMD861byK OXUjJptsU3H96GwFb8KxbpRBqojibgt09KmSsUtTql8ySJoy2PeufvYY1k8tgT3wa6C2J8vmsd0U 6kyz9OlYc7ub8isZv+jxDzJAiof4npbfxBZzTm2trhZJRnCQpuLVe1KxWbTrqzeJGmuUZPtG7nGc jA7Y9q5jStGvdP1vSZ9Qmh8jTLXyIBbR4ZgP73r97P59MnPQoq2pzybWyNk6hv4V1dTwSvBU+4qp cL5s8cvcGotYie71Vby0XynJ+dv7w9CK0VgbYN4wetYVHbQ1jdnLapZf2h4rt4GTdGELOB/n1rCu LfyrqWOPLqjYVvUV3Kj7LLqeqMvzRxARk9yB0/WuJZWy3zsTk124G5z4myK+1/7rY6dKdhh1DD3x /n2qdg2/BkAz6ijqcb0Oe1elc5LkK8mn/wDAqeBkH/Vk+460/Gz5R5bD1p3FcwQwAx+vvTpJt0RX uelRSAruHQe9QyFccH5a+a5bm9xueufvD0p6SHnIyvpUDiHPG4t1zmpbKSNAzXG8qSQu3qK3tpYC VlBRiMcdPU1k3JO0qn6VvtaGWJpLC4DKP731/SueuN0cro68iqhBoaZ6X8IZZb/R7yzf5xYS74h3 /edR/wCOn8639Xto7XxBAYv+W0Bbpjb0+WuU+DV2ba71boN8cfT1Bb/Guv1+eWfUYrmdSC3CZHO3 aP65ofU6Fsi/ZnJbrjOa0Db+aCP4aytNbduHcVuwFcKCazRqjPWzKP03fWrKRtx8uAetaIVZDmjy 8Z29MVotA5SNz5f+9iuanOb7zM9DXRXOQj+p4rnMM9wqJ95uSaJSuLlOrtcFc1k3Z/01jjvWvYxY gIJ6VlajH5dxz901zTOiLLzwrJFz+YrNa2l3YXB+ta9oMxAdeKUx44PNaKbsTyIxfsLF8uc47AVJ MAqHNXn4OAP85rOvnxGW6Coki0rFfVY4f+EPY/8ALRkJb/vqvMESPZnzMV6Pq6PD4aO7nzIcgele c7/lYfL6bq9LAu9zzsWthSi5P7zOM+9Hl4b5GB96CHz95OtJlg7ZCnsfSvRSucvoP8tv8tS7JD2/ Wmb+fup69KQ5zwBTsIy4ZPObcfl/xq1dWQvoF+zri5BxjpuqgAuQvVc1ZileEApLyOQfQ+tfNX5X obaGbPbyWc8kDxt50btE8Z6qwbB/UVpSRQxgQRLuI+9/tHuageHzLsXgY4DMzLnPzdv8aeJMPn0q 5u60AkjWS2fdFuA6cnNZ1zCfMO/BzWpE+VOAd1Zk8x34HPappymNNG14ALxa5Labflurcj8VIYGv SvFWJEhnTpG4H4V5FpWoyaZfRXsO0mI/cPcdxXoc/i7RrvRrr55BK0R/dFfmU1stXqbRkuU1rB8M x9a1YpTnOawbB9yBgQcitS33Ecc81KOmBuQTZHB7VZRwfpWXB/eHB7VfjP5VdytCDU5vJtpXUfMB kVjaYqCf5j83qa270I6NG/3SCK52aIoCm5i3AV1+8tJK5m5WO7s7TMGdwGfWsPXAsR+Y5PSqMWtT pCscsbbl67eh96p3E0t2ys/c9v4R7U5xTQozaZv6XJ+52H+E4BqzPJgn9azrPbGmE7frVmWTOaxW huV7iXANY17NuikUfnV286Max9Qk+z2zN1Yc0twctDT17b/YrxOcBItorzAeUR1wxP3h+lamo61d 375nkITklQpxWWRCRtzg88ivWwlCVK9zzsRVjO1g8uPnbIKZjGfUHtUpjTOFl5P+f8Ki7iu5HMKB zj8/8/lS9ab7fhS/jTKRlCTg7SBTm3fMM+1NOdx2tSAHBJYCvmLF3LcOTDIAeR0/z+NV0OGJwpqz ZDiVcgjrVbIwQVYfSotuIswFfbkdqyriPMrBKt8h8qSRmoZw3mluSDWkd7hcqlCvekQ4O4VO4z1/ GoGBB+btWqdyj0jwpei60yGTqU+R/wAK6+z2565FeWeCb3ybuW1f7kvKezCvSLGYFMURVjqhK6Np hsj3VLvG2qwbzLfbnNSIT025YfrV2KuRylnJDZFMh08ySZPNUdRur60Jby0Oe9GnzLeQE3OoyRSF tvlquNtUosn1NuTR1kfoSAO1LLp6IgTcoYDkZqy2hW0c2FvrpFbjb5lZeradp9jki9mB25BL571b g2SpwQ8xeXuxxQshwB+dc/p8moXM7R+exjz9729a3JY2hTy92T61zuFupvzEcg+Va5rXzlHQHHGf pW5d3KxhmzwOK5We7Fyl0VI3NwgzV4enzT1Mq1Xkjcx/nGVE43ehFVmEifM8S7SP7uRVoLJvY4bB 46U3DHh45Bjute5E8u5X8wA8wqO1R+uOP6VamjLAjdJ36qear7XB+41aoaaE79KDQARwRilz7Uxp mPxnBDcU/wCQchTVmaAA5B/+tUWQOPLUEDvXyXMdA+yx5xCZHFMKxbm3bhT7WQLc52gH2pSIt7bi w57UrgJ5cJJ/eH8alW3XBKtxR9lDA+VIsmB06Go4d8bEcj2ptjRXuYBDKRjj+Gqnk8/MK3TGswcZ /ew/fX0Xsfp2rOuNsYJNEZNBYTRWVNYtt4+Rn2tjtnvXotrI1vcGGX7w+6394Vzui6UkekMXX/Sb tM7j/COqiumij/tKyiflX2gqR1DV1W0uaQNq2uMdOhrRt5QfrXK2Vy0btbXIxKv5N71rWkxL4B+a m9TRM1b2NJ4mVvTg1z32QRy7ZMjn7610afvVYAc1VuLQn+GhGsZWIBlQNl+wHT5jzVK6WDOVLTP0 G7oKfJY3G44VsGrdvpr9XHSqkUqqWyH6XDsPmfjSX90o4XtxViZhbREDiuX1K5ZnaOL7x6VjYiUm Nuh/aAliMrQxYxvTr+FZ/wDwjoA+TVZ/+BIv+FWruMwQxKuRg4NJBckDbKOa6KV1sctWz3Kv9gXQ zs1SM/78A/xpP7E1LteWbfWMj/2atdTkbhTvSt+aXcytHsYn9j6z2Nk//A2WmnS9bH/LtbH/AHZz /wDE1vU7J9SKOafcXLHsc4bDWh/zDwf92YUz7Nq466TN/wB9r/jXUb2z95h+NHnN/eb86fPPuLki ebu7b25+/wDpULiNN3G4+p6CqllN5kIjJy0fT3WrQIkDHqfSvHcOV2NrjUb9+OOcdambbubcrDry tQ879xGBk1KSQ2OuaCQMZxuRt6+op8UhI+cHp+IpgyGDIcVccxw2xlK5HpUN3GkYWpXssN950DYk jTZ7MD1H0q/pFm2vLLMSYYF+VmPOW/uisC58y5kbyxkyH5V9T2r0/T7GHTLK2sU+7Gg3N6sfvH88 16EKaaSFcncYC44wvSr2lyAOyYwp+ZPY1UlQhueR60yMmN1xwRWnIrWGpNG/fafHfpndtnT7jisg XNxYXKreLtYcbuzVvafMtzFkcHpVuezgvbfyrmMOuMfSs/I2t1RHZ3qnDA4/rWutwrL71xlxpd9p Upa0zcW/dT94VJDrQHyyBoz6OMVVgU2djJJGCOhqKW7VItx6deK5r+1om5LjNQvf3FwNttE8n4cU mrl85Z1jUgqNsOGfgVnaXZmVzI4O7+VXbLRZpZPMumyx/StwWqRRYTpWUmug4ps5LV/llXjp+lVc A575H41F43a8tbqC8slJ2bhIuM5X6UzTpkvrQTQ4EinDj0NdFHWFznraTLNvLImQy9Dj61eR1YZ6 VUA8xMp/rP4h61JGQpyR1qyCzx+NLmoy67tp4b0p+PaquhWHZ9qVTxzTOtJgGmI8ehPyqyHDD7pr VspFkYsBg9GX+7WPZn/R1PtUvnNbyiRASR1X+8K5KkebQpG2ybjnqT7U4rnClflPT2qW22vhkbdG 3Kn/AD+FPCghR7c1xNNaMqxT2kHafoaratMfKWAc4HzfWtiG2e7kaK2+9/HIR8qD3/wrfstCitIi 0EW6Y9Znxk/4VvRpuTuS9DnPCehy/b7a/uQsccbbkRvvMdpwcfrXVPG2wf7OVNSMgj2yu3zBs8/e /wA9adM0azEt0kArsUUhXY1SGRd3ahvvZ6GmfedicjHWpJiu/dnqKoRNY3f2KUsBlT1HrXX2kiTR LJEcqa4pCOSe1W9L1NrKXeWJQ/fX1H+NRKJpGdjsgOMYqJ7O3mOWiXP0q1ayRXUXmwOHWpimO3Jr M3M7+zbcZ2xrx7Cp47Re44q3gDccUjYAaoZUSHCgkdqhmXjIqyikk9qjuRhayaNUzkdcjDy+uRmu JBXTNVlixiKTb/wEMePyOfwrvNRyblSOQK4rxVEEv8o3BRfw610YV+9ynPiEmrm6i/I0qnDA9vSr Kqp2sO+CRVKz3FFDjG0cVaO5ZVb/AJZt97P862ZzodOMAsAGI5zTBcj+JBip9rfeVfmX761UvYdm 1gMK3NQUWlkiJPp7Ucf3hWchbODU2fenzMmyZ5FZ7o5JIHUqyk/K3Vat+WS2e9dn4s0CfUPFEE9o qrEEHnSno3t7mpIPDVpFgyNJMwGP7oqZJvYVjndJufsreUylvLfcqgckH+H8607PTtQu3WMwNbQH 7zn72PQe9b0FvHaJttYRHgE/KvP/ANetCNFhVWySx53HtU+wV9QuFna29jAI449gVeh7ZP8AOiaZ sYTj2oLjJLdM02UdcitkrCuMWTcMNzz3pSN9pIp+9Cw2t04PT+tMxlfSpIn+dVzgyKU/qP5UwKxx IFkHf1zQzEA98/rTmyFmUKT82Ru9+aRehyPy9aACB1yS56jjNWV8sleBt9+59KqcCTlcgVf0mLec vyq/MfzoAsW9zd6ZIs0Um1ifnjPIYe9dvp19Bewq+5Vc/eXNea3kkxa6KbTLIhETP0z2BrMsNfub FLYzkFi6xyRuMMrHt/8AXqZQLjM9qaHqO/Wqssb5wvSuesfELqijcSo6q5yfzrpbW4juoUlhbKvW bVtzaMk9hI48Dmqt+cIx71qYBHTOay9ReOGNmldUA7sazkjSDuYTW24MW/CuA1grc6wEjOcuBn2F dJ4h8S26xtbWDb2PDP6f5xXK6QPMvZJTyIxjn1Nb4elJLnZlWmtkbpIGF5xntV1Y5Gjwv3e+7pVa zhEs6huh5eobqb7TMZCvThfYelHLcycrGvbo2wsjxS9tyvnNPnCSWqxzHqD8391q5hIIbTzJBCoX OW2r/wCPV0q5FmpUbsdvwqnCxPNcyJ0KSkN96kwavzj7T/Eok6KfX2qoySIdrIc1Nhk+Wb5UHsvN K1owBMr4Y9vSpXMcY2xjuabvyfm+9VCAYA7YqM9MbfpQckNxikHD4xSAXjPSgAF1UilHQnrSuPvD uKdwIcde9NY42NjkOv8APH9afk7z7morr/USc8iM/hTQhZ0SJ7lWOPkB+76HAqL5QPlkqxdW1xJ+ /h25dMc/hVVLS4aNGhj3HkfN/DU3GE4YxKyJ1AO3vndWuQtlbfZ1bLsxJP40mIbRpPL5lKY9lxVZ iWwx+9VIQrRoQeB9DVXyoQ6NNAsoT++PmFWgcn5etKACCHHequTYj8tHL+TKGP8AdcflU0V3f2m7 c2M/kahkiw7FeHzirCXZUbLlA6GjR7j1WxfXXrySFrf7WYJDwkmM8/jXFa7/AGtHORqkssmfuPu4 auynsYpQTEMDOBVHyw0Zs76PfAx7/wANEOWLvYpzl3OIRPMl2btuf4z2roYYbSI7LF98W37+Pvf5 NZWqafPYTvb8mMj5GA6j3rotMtUjgZnGUh4A/vVtVknHQlFoqLW0PTz5xn/dWqiRJnaeP8ae0jSS tI/U803LE8fSufYp6kNxGwUcZzxj3q3ot7FDH9ju8mI8I/t2H4ZoI8xXXHzDkVAI1Jp3JsbLW1sN zG6jK+uafHc2sahftO7HesPy8dRzTAhHHSpsUXD3ycGgDcQMdKRdvVuQfWlPt174pDBfm68U/Gd2 OlM757HsKeDycUACqd1O4wcUvbnpUfQ9vekAjqQW/Oq0iiRWVuQRg1bY5BFRY5yPSquIkW6nWIRH kDvjmm+ZKQOeMdqTpgfrTgPl+nSndAMA4IHbpS4/DuDTu5x2pQCR0IpCIlG1jnv3qZehz0P6Uxxn cCtLn+LvmgB/ba3Wmug+bH/fNOBGMik5HWi47CW0xh3RknaelXAFkt9r/P61V2gg7wPQU+IlQcnA ovcLCXwdohbqcyzt5e7HOKZcsmz7PGfljOG96lU/8TGEKCVjhkf+VZ1lObmAK/EhB4Yc5obCxI2P mxSoPehwdzEdqdtGR/KgQg4b7pPFMIAdsU5sfN+dM5IOetSUTEqVHHXtTN2ewo6g0hC5oAe/Uleh oXpg0vHC96UKMe1UIfj161GDg7ven9AefamN0xTESMT8yr2weKjzjnotKGyhI45zTyOcdqQAF+Vh 6U3GMqB6YqRe2PpTWGDkZNIobhQcYp/zY5prDnd296cnOd1ABtAdu1HJA4x3pzjHPoeKPXmmAjZI yKRgp+UcU5+9M4zkdaLisJGcKA3uKe/XIqMdWx065qRTvWgBoIx0/wDrU4Hn/ZPagY57MaftyFpD GRZEynnIjkC/pxVKNJ4tRkmbcIpBiNccr3P86uMD5iMOG+b8OKR5nifhy/fBH60MELKM/vNrDIqL PPI+tS+Yu50O52PQjpio2GKAI075FKORwKdng9KE2gdOM/lSAYeOtN/CnsMN3wTTG69qAJpRwccU iMTz+XtRRVkju5pH4x/KiigCPO3cO3Spx93PeiigBc8n60A5DA/54ooqShue340q9z70UUAPY8kU g5A9ziiimIbmjHFFFIY5htPHT0pF+V9vUUUUyR+PmIpR/d7UUUigdRjd3zmoJZGZFwcZBFFFNiIR mRFBPTNXXUY4yKKKQymR19qTPBoooAd2xTMH1oopAf/Z ------=_NextPart_000_0000_01C85294.643E2190 Content-Type: image/jpeg Content-Transfer-Encoding: base64 Content-Location: http://infolab.stanford.edu/~backrub/larry.jpg /9j/4AAQSkZJRgABAQEBLAEsAAD/2wBDAAkGBwgHBgkIBwgKCgkLDRYPDQwMDRsUFRAWIB0iIiAd Hx8kKDQsJCYxJx8fLT0tMTU3Ojo6Iys/RD84QzQ5Ojf/2wBDAQoKCg0MDRoPDxo3JR8lNzc3Nzc3 Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzc3Nzf/wAARCADaAKMDASIA AhEBAxEB/8QAHwAAAQUBAQEBAQEAAAAAAAAAAAECAwQFBgcICQoL/8QAtRAAAgEDAwIEAwUFBAQA AAF9AQIDAAQRBRIhMUEGE1FhByJxFDKBkaEII0KxwRVS0fAkM2JyggkKFhcYGRolJicoKSo0NTY3 ODk6Q0RFRkdISUpTVFVWV1hZWmNkZWZnaGlqc3R1dnd4eXqDhIWGh4iJipKTlJWWl5iZmqKjpKWm p6ipqrKztLW2t7i5usLDxMXGx8jJytLT1NXW19jZ2uHi4+Tl5ufo6erx8vP09fb3+Pn6/8QAHwEA AwEBAQEBAQEBAQAAAAAAAAECAwQFBgcICQoL/8QAtREAAgECBAQDBAcFBAQAAQJ3AAECAxEEBSEx BhJBUQdhcRMiMoEIFEKRobHBCSMzUvAVYnLRChYkNOEl8RcYGRomJygpKjU2Nzg5OkNERUZHSElK U1RVVldYWVpjZGVmZ2hpanN0dXZ3eHl6goOEhYaHiImKkpOUlZaXmJmaoqOkpaanqKmqsrO0tba3 uLm6wsPExcbHyMnK0tPU1dbX2Nna4uPk5ebn6Onq8vP09fb3+Pn6/9oADAMBAAIRAxEAPwDrDTHp xNNY1wHUQuKiapnqFutMRGabTmphNUSL0p61FmpFP+NDGaC/cX6CjvQv3F9SKOvQE/QUAFITzSgH 0PXFZ+r6xZaPGDeyHzH+5Cgy7/4D3NMVy82fSgg56cVwF/401GUn7FHHbr2/iYfj/wDWrDm1TUbl ybi+nbPUeYR+lHOiHI9akwmN7Ko9WIFRJNFNv8mVJNhAYowIB9K8i+05bcxZiO7Guo8AXjf2ldWu cpLbmX6MpH9G/ShTv0FzHbOyqm5mAT1Nc74ku7t9JlfShmFpwjS20u5nU53ZXHc54B71o6xc3EUL rBZLLDtKyO7foBj9a5eyuYNCimgtEmhF3ibEvzY2jlB0ABySH59McU3JLQbZUFzc6LaxKLRDezyH zZJvnLRFQBGB/DjHPPWs/NpZypb3Nu8yhwXkibkAH5gBzngdak1C4S7uTIqImTxHHk4GOMkkkn37 59qteHzcrfvJpEkTTk5WBlBDKQeRn3rJNNknQx6P4Xvo0uleCNZlDhC+CuR0NFcfJN4faRzqMV99 r3HzsSYG7PPAGKK05l2DXueuE01qU8c9qa2fSo0Nxj9KhapWI9QPrUTA4zQBE/X2phNOfnntio2I HJIFUSLnvRJPFbxmW4kSKPpuc4FY+s63DYRsqMDNjof4Pc1xl3r8szs02ZXbjDD7v+H4fnVxhcV7 Ha3Xja3UFbGzknK/IrSHYGI74xnH5VizeIbyV3e+u3iJ6RQ8BT6VzR1KWJSiBdzH7x6/h7VYjupU HmTWTynGDNnIH6Yq+XTQVy5JqsruqrLOxPPzNjNQXDm6bdGxDbfm3cn6ZqYLBOqNbOW/veYBgcUX LQRyP5aq+R97OMH8qlx7jRkTSSowDjpxjpSB9yls8g1LdrLJsZipG3GR1qlLugbr+VLlTJlEnJJB yK7D4dWrf2ndXZ/1cVsY84/iZgf5Kfzrj7d0lYA5J44HevVPD2mnSdIjt5B+/kPmz/75GMfgAKmz UrEpEHieTMdnZEgfaZDvlYErFGBljtHXNcpq1vptnbCWxvBPIH2mFlIO36Htya9Abbv81sAKpyx4 ABOTXDeJtXsb+7hihjleOIt+9UYDueB7kdKKkU9x3KltefZoYGto4Y7kbkjmHO1SOSw7nnr2xVGD ZE/mwNtK/KMZ4FdLoOi6ddaRa3U0PmSSKWY72HOSMYH0rSTQtLTIFmpycnLE/wBaj2cmKxxdxaWE 8zStPLCzYLIoyAcc/rzRXbHRNNJz9jj/AF/xoqvZzHYs3nivToCRD5lwRxlOFP41g3Hi+9kciJIY lPIwu4j8aw+CmD1PeoW67TznvXM5NjcmbK+K9UjYFpw4PZkGP0FTXviKbU9Pa1ULbytIpd1P30HJ XHbJxz7dK50+nQd/amj5OQBke9GvcV2dv4dvGbT5Yrtws6XDbQWzvQgHP57v0qDXNUFpaPNjkkCN fVu1cs8peMCFm81yEjAPO7t9Oam8TXiXGrm2Q5itCVyDwXPU1VNybsaJqxjXs8ksxJbex+9xxu71 UO95lRmCjtk4qS/uBvZYflA/n61R+Z3BBAPTOc13xWhm2SyLIrNht5X0NbemX6T2rQzMkZA4I53C q0EoitkGWVX+VsHJb6en1rc0XRRqZD2Ec20AjzXOFx6VWiErsiSFo2VkuC0YGFZVyPxHale2mZNo jMpPG9V213ej+FYLaJN482RDwW6D6CtoaaFXlB9MVlKSNowkeRT6XdN0hdSP7xxWPd28kbESfL7G vY9W0+NYw86EoOMZIxXnuuQfaZWjtLFgin76oTn8cVMWhygzlre48i4R0HzIc9K7NvE95d7JIp2j ZByB3PrXN3OmXFtG0kkLhVGSSOlQ25dQNp+fBZff2py1Whk4s6SfVb+6iMc9zI8bn50J4NNgukgg niS3iLSDAkIyy/Q1RhlEkCOv8fP09qkTk8jpXG27k2Oy8Njb4f09QScRkZP+8a0QQe4GOvNZvh// AJF6xGedh/8AQjV/xYi2ng6XVoUUXUEcW0np80iryP8AgVd1PWKB6DywBwWUH0JorG+xakQCb+AE gEg2uf8A2aiuj2MjL20O5yhl+UfKevrTDIGJ+X9a1H8P6gnAgLc9VIpq+H9TY/LZt65LAV5PLboa 2ZmGRckbB1pGlGT+7XB9RVrU9OuNNWM3gWMSEqvzZyRziqLvGCRvU/jTt5AKlyYHE42q0ZypA79q zppJjkqB8xznuTUt5IgRFHzDOeO9W7C1LNGrHMsj/KvoOK6qMElcavsJp/hq5vmRpGCq5/Guws/B FlCgzkv0ZjzgcdPStbTbdLeGNsfNtArXQ5xk4pSmzojTXUzj4WsJo0iijWOPgMcZYj6//rrpLaxt raKKGKNY4UGFjA4+pqO18vA5Bwa0FAxmoc5GqhEkQLkgLnFOcegpm452ij5mOTUORqokNxEJYysg B5rFubJc4HTsMdK6FlyuScVl3i/NxWfNJFOKZzGsWEf2WVSuVdGUj+VeawQbYnX+O2kwD6ivXNQT MLA9+leX6pE1rqd5GFJyN4HrW9J33OWtG2wy3jMSOhJC7yRx2IFWFGOByR+X59q0vDljb6rBczSN hYWUcvtXBHUn060moXmj27yWybHl2MEmg+YKR0x2Oc9fbvUSpS5mczsdDoP/ACA7If8ATM/+hGrf jR8/Dy9Xt5duP/I8dUvD0hk8P2EpJJaHOT1+8aueMRnwDernqLcf+R0rspqySJew+c4kxz90d/YU VDcXNv5p3TKDgccegor0TzdS1kfrUsbDbz2qvmlLrHHI7nCBct+FeW0enc4fx5cPc6ra2cXzeTF5 rKP77nv/AMBCn8a594Uj4kkO/wBVHArYuD9ou7m6kxuuJSxyeg6AfQACqVwgLc4IP41zSqXlZbCZ nXbKjbRyFxg10HhANdXpmf8AgAVfbnNc5dQM02VGE/vGuv8AAaYjuMjmOQBvY46V0p/uyoK7Ol1K 9ks4FFtH5lwxxGpOAOOpNYkcer3kpkkkcMTghf8ACulW1DjzW6EDOfSq66kv21LGyTzbp87VXqQO vt+tTBG8jCaXV7N8GO6GD13ZFbOh+IdReaNbtXULKFYN6Edf0pun+JBqckFpb28jXE1w8AhBG8FF BLFc8LzgH1Bq4JlkVpBtYI2GXGGUjsa0ZCvc6621SC4hRhkEjg+tVb3XooI3CAFgucGoNLjhnt4y q42sTxWRqyxPcSIiZxncScAVmnG5u+axRvvG/mR4hKg9ye1VIfE19MpcKroOCQMYNWIIrZry1gjs vNmu3ZYQkeQ5VdzcnA4Az1pbbVdOvbeN4Yl8p87H8srux14I5/DNW4prYyvK+4208QJeTJbXK+Vc Ftq56PXN+LALbVHdxgmIgc9ea6ibTopSk8aDKNvVsVy3j7/XK3YxsayjFKehUm3HU5lLmUWzQK5V GwWUHhsetVnDbSQMY/pTrEh7kI/A20+eKRSU7YJz7Vq7JnIemeGs/wDCMaXn/n2B/U1f8X8+B7kD Jz9nHX1nSqHhvjw1pg/6dgP1NX/Ff/InTj/btv8A0claIfRmDcPKZm2hcdOgopgkAzkZO48596K7 jgsdLsNZHiK9jjs3tY5VMr43qp5Qe9YWseI7i93R2TtBa5IDDh5Pr6A+n/6hn2ahLX5efMYtk+nb +Wfxrx61Sy0O8WUr65XPp0qpKCr5B49qtyHpnr71WmBYgL1PrXNAlm54U02DWdO1mzmyJGa28p8Z 2Hc+WH4DGPQ1q+HNJuNIlvIb4J5pkVyEbIxiq3w8uFiur9FI4SOUZHXaxDfoRXRXbOb92lBDSqrD /dyQD+ldEW+Vo6YRVky2kXn2yxFtoI5qE6bbQ5CRrg9cdfzq5bLuwOmK0ktgwAHJNJN9DoUE9zFS 2hSMqsCLk9QOfzpssLGJwkYUd8VvyQW8I/eH5vSsy8ufNDrAn7sEru9abc+o+WK2JtHfy7YR9hzV TyFleY4BLE9Rnirulwny8gHaeM1Hn7LMxC715zWT5kzZJOJnS2FtIvl3FpFLGAQEcZAz6elRto1p LAluIAIkOVUcBCeuMYxXSKtrMgKnk077KgPyjitHOdtSFSi3oYcVl9liKByyYOM8kcV598RSVa1U fxRFffJYV6feoYxj+H/GsFNHsdVuIbi7i3ywybY+eOCOf50qcnzGdWnZWR5faaRetbC7jhIXzNjS OQqrwDgk9+elWry1txE7PdrJKoxsgUsg9y5x69gfrVi/1yaXVb12lkW3N5I48rAIXdgY7dAK1dT1 nS73SJIkuGErKMAwhD9Bjp0611Wvqee1Z2N3w+MeHNNA7Wy/zNXPFZ/4pGb3ktv/AEclVNBXZ4e0 5c5xbgZ/OrXi048JSf8AXW2H/kVf8KpbJjOfUZyf9o/zopsUgK5/2j/Oiu65yWOVncrGxVeFBP41 rSQiBfLzkxARn3KjGf0rb1bT7RNKvZIrdA6W8jJgdwuayb3AaQg5BckH1GeteJiIuNjrRRYhnwOp 6CopCCCAeO5qcxEKGxy3GaJLclGxxgE59TWSaFqXPCl1FZ67A1wdkUyNCx7ZbGP5V3eq5+1WjErt MTJgEE4GMfzrzSSMRIhldVwOc9TWh4Vvg+sxW5dmUxsELHgdOAPoDWkZM3py0sz0m0bHB5wxq7Nf Lbxkn73YZqjaHdMc9c9KrX+fNZn4GcLW8I6GjqWJpLoMC5OXb1PQelUHmmAkVSFUkso9/SmNNEjL 5sqg/Wpo7q0OfvH/AGgOlXYE29jQ07WZLW2VHhPI+YkVV+2zySlyAMnAUc4WpzJY+WHaUiMnG0jn OKr+daDPlyqCW6GhxjYa50WQ4RcqcelXbO93oVb7w96x55ECERurD2am2bF5k8o7snBpWugdRpmn qE4IJ9ASfwrO1m7j0fwpLeH5ZEgKJ6l3Jx/P9KsXAB+902kn868q1jWb3WAFupy0COTEg6AZOPrx WKjqOdWy1MpwrKAfvevvTLkH7K45yBwQanx1xQYDJC65xkda3TOFs9I8PjHhzTB/06p/KrPi8/8A FKn3nth/5EFV9C40DTB/06R/yqXxgf8AimkX1urYf+PitRrY56AHy+nc/wAzRUtoMwL9T/M0V22O W7Nt41bKsNysCpHqOh/nXIKpiOyU/LbMYCxwAxQlf5AVu6xqUkGbe02ibHzyNyE9h79DXMaleS3Z jt7h0/dlmxgAMSc5Pqea8qtyzfKdKdh82oWsAYOxdt+dqDPGPWqNzq8sg2wIsanv1P51WnVkRWGM lyAR6f4VFtA+p61MaUEO42ZmZ97szOe5PSrOkTm2v4bkAnyXDNj0PB/TNVZOEx1qezDfZZXUEEnb +AGauVktBxPZLKVXKSK2cqCCKu30KMYnIB9c965Lw1ctFZWttM3z+Sjox7qR/TH8q6r7SJYQf4um M047GtzntYtPs0guYYlkj3/OpODjIzg1s2s1hNApE8QfyhlJMAkq21jj34IHvTblFkT5hlD94VmR WkYk5RZY+eD1xVNs3p8rO4bRbOS4ihRUMYU5fIzn/OK5zV4rK1jmEkkbMqFxHGMs2ThQPyP0qV7P TVgimt5CZMHcuMbT1/Gs6SGI4SGJY09QPmNS+ZGqgnuyto2m+eyXFzEI1RjtTdkn610FpEkRncqM 5yMdqigAjhyowF4X2HHFVri88qNlBGXoTbRz1LJ6FDxJfCz0q7lU4ZkMUQz/ABN0/wA+1eYrDtCo eMDBrf8AEeq/b7tYov8AVW7EDnhm7msZh8u08jHSsuazMKkrlcoVz1PpgU3eyhsg9D/Kql6zW106 ozKD8w57VGbyQRvlmPBHX2raMW9TM9Y0b/kB6b/16R/+g0vjI48PQj/p8tv/AEKk0bjQtM9fskX/ AKAKb42OPD9t/wBf1uP1P+Fa9CkUNPTNohx6/wAzRVrSot1hEc/3v/QjRXctjkuco1xK7biwLliT k1l3TB53Z1z83c1clVZFBO7BPOOKheIM4bON3bFeHDV3OoqbCWy2T6DPQUYGcE8jt1qa6YKTGnGB yahslaW7ii5Ks4BrdbCJ5baCCBZJmeR2OBGFwB75qq9yIoWSMbRkkjr2rQ1eMkFl4RGyAPQ1T0+z a7u4oVH7veGkbHat1Fcl2EVdnfNaSpp1m0I/0i3ijC+5CjIPsat2WpCWAOM9eQeoPoaskEqB6/rW JfQyWcrTxAtE5/er0x/tCso6m0lY66B1lgGRUMtluYmI4LVl6LqKMgiZxuXo2fvL6/0/CuktWVwr KwIJ/wAmqs0NS6FAWFzs4YgbsVYgsfL+aU5x2rTby8Ad854qC6lRFIB5od2VdFS/lEUGQOPSuM8Q ahJ9huHiYjbhcj3P8+a0de1NmRLO1bdNK5GB/APU1znieb+ztPsrZMMZJWLbupAHJP51FmtCXqYg YKMACnoQx5PP1pnDYIwAeenT2odCFyP0rGSs7MxZFd2a3d7YKz7fMmEDvjO0EjB/8e/Sma34a1DS 4Z5ZAkkEZI8xG684re8Lxpc3VzbS7STEsiEjnKuDkf59K1fGZz4avW7Eqf1FddFe5qQ73NbSMDRN M/684R/44Kb41BOh2uFZgL+BmwM4ADZNP0zjSNPH/TpD/wCgCrgnkHAY49KbNInM2evW9rbJDujO 0dSzA8nPTbRXSGVj1VD/AMAX/CitvbSMvZRODjtzJkFunPWkmiWPc3UJV0QjZv7EdQarXMDyxMVI WLdh3I9686MXYsxnO5mOOSc1fsLVreKS7kTDL8qKRzz1P5E/nViKFLfHy+Xu6MwycVM8yGKQgfKO eTya6o0nuyXIy71jIzxjgKvT1NdBo0dq2nxfZVCtuAlU9Qf8K5hN3O7ls96s2l3NaTrcW5G9W+ZT /EPTFXUjdaG9G3U9L2gheehpksSsjq+CjVFoepW+r2rTWpw6H95EeqnA/Me9aQiBi45yKwinHRlS 1OTu9Pkt2MtqMgnJQHBB9V9KkttaubbKyZyOPmGCP8a6GSH5uneoDYxvw6ggdvWtUZO5RHiSQr90 Z/3qqXOp396CsI2j19K2hpcP/PMDn0FOe1SPooFKTsUkzF0+wEBaVvnkbgua5Xxg5n1uGBTkRQfq Sc/yFd3NxGccf0rhr+P7VrV1cFgI+FBx6VEE3MqfuxKkcLgBACwC5J96JQUHII9jUluZZTcNECVR No9zSLL/AAzKSD6irqUnLVGKs0U5JbqKSO6s96tAd29QSF+vseRXVeI72PUfBUl9EuxblVbb/dO7 BH4EVj2N7Np0UiW9vb3O9i22ZScggDbjPIra0a2huPDa2N6YfKZ2ZY42IKgnO05zggk/hinTXIrS Bx7G7phDaTYFTkfZYhx7KKsYYc7T+VYEfh23iQC0vr6FBwoSbgfpVLXIdX0ewN5ZanNNGjfvBKAS o7GnvqgTWx1fB7/pRXma+MNbx/x+L/35WikO6Oz1NLN7nzdOsFUBsbVbDN/tbaz729jssFgLi5xl VzlIee+Op9qyLrUJZ1C5CJnJVO59z3qFWZ49sYOQa0hTURRXNuJdzyzEvMxZ+c5p9s7yQDapOOM0 R2pYZmP4elXIGEUgHAQjBrRrsZORTmgYgsowy9QaqFmUs6gZHUH0rddN/KI7rnIdVJFUrq1DpuKs jEHkjGaNxqpZD7KdtMkiu7SQpKWCgjo59CO4r0jw/qcWr2pc4juYjtnhPBVvp2FeXCIm1WKckKWL I69UYcZH5fpWhe6vNc3UN0VEV3HEIXmg+XzMfdZh/e55Pf8ACs50+Y0jWsz1GSDOcj5hxUEibRg8 kdDWb4X8U2t7p4j1O4SG7i4ZnIAcY4P+fStC41PSSuTqNpyOolFZarRm9k9USKVK5JGaoXM4kYhc YAyTVC81vRo85vw5B6RgnNYWpeKIZrWa20+3kXzF2NK5wffA/wDr0cvMxc3KRap4iMzNDZoDGCRv J69qxR5k52g4UsQ2O3v+tVoiIuCDs6D2rW0yCF9KvLp5NksUibUP/LQHGfyxmtlFRWm5zzk3uWIb d4AIxE+5eGPY1WuE2MY5I2HcEitS+ijE8LRvlZYBIRnjduZePwUfnVC7OJE69Ki7JTXYgNvBIqnA 468mk8pUO1XZVJ7HvUyMHzvVTjvjBpjInOJJEB9QG/wokr6NGj2umbulXhihaG6fcfM3LJ2IIAx+ lWNbjS50a7g8wDzk2q2Mjdnj9cVjwAtEr+fz05T/AOvVpfMFv5qMjNBICQMjOSBnHqCR+ZqU7KyJ 1buzzmRTbSPBcKUlRiGUnoc0V1t9o0F9dSXVwbt5ZDlmVlx06Dj8KKXMirEEVtGgwVJI4OfWp02x A/wj6VYuoEQ+YWd1LYOPlHsf84quFC8pGgOepGSPzrZNvSxmn1bAkM2EVnP+wMj8TQ0Cbc3JDEDi FG/LLf0qTczqVZsDryxx+VVpJo0YAkcelVrsyml8UUaMczXMWJHwy8YXhQPYdqaVRMpcZaBjk46q fUVmi+EbAopb1zxmrCuZcMz8Y5/wpryOead7svxxxXU8ds0iqkjqu/8A2TgZpnjPR4tDvY7W0mE5 2iUv1wp42n8ifxqKJImZY2O3cwKOByje3rkdvb3p1+LiG9kjuRudWOXHIdeoI/2cGjW+rLjFWMSJ 2zuIIyecdKCCXJK7Qew6VZuLZl/fRHaD94DsaYkgGRMAeeopvlNowkV2ztz+lTxRAgEE4ByfrSNH uuIkj+cuQFUDkknAA98mrN1bXFjNJbXUbRvGSCpFQl2NY8ql7xXADyKg6FqvCMCJkbhfMDDH0xio LMAzZABCgmr0SrPdQIe8vIzjtnH6UOT7ESim9GFznyraEZDxR7HPuf8AOfxrPnuJlueTkLxWneZN 7csxGfObgdODisu5U/aD70k47k+ylfQmivByGQ/UUi3kXQ7uvpUKAhugOaGQhz0/Ki6Zq6L5L2L1 pfRbCuW+96Vp6dPDNdNb5J89Sn0zwD+tYVoD5jDHYmtCzPl3cDr1WQHp6c4/Sk43OdxcXsOjvLdk Ul2UkcgdqKjukEF5cw/LiOeRRx2DHFFRysfKW5VV4SHKgN1PpWNJdBHKAZIPJzVtcyhXZ94YZ9qq X0AVhKvAY4P1rZXM4pERZ3O5nPB6e1PeNeWwMA/jTEGQP61YjCkFT19T/n/OKttI7KNNyvHoV9ik EY71LZIm/wAt8kE8U1mwMY6VEzlXVh2OetJ3epLjCOj1ZpeVGE2szdc8fp/j+Aq/J/pWnfaA2bm0 2q+e6F8Z9sM6H6Mf7tURKroG7MOav6I3m34tiv7u7XyTx2b91/KXP/ARSsjndRrRFEhdpOMKT+lZ lyscMzJ1XPFakFvLJErSNtLKN3scc/rVbULZEkU8sSuM/nQmSpMj0xWjniu4CFmt7lJI2IzhkIYc emTW5421RNb1SK7trdolWEqyse5Ofx+vvWTaFI4WRiB827n6Cp3aIx43qMHjBqZJXubQu4soWO8S tuXA2n+dWkJWaEgYIlUj88f1qW3IEw2sCGGPpVyRQOcZwwPTpgii5jKWpU1JVTUrxVzhZMj8QG/r WdeHEqHOMj+tdHqluhv52ZBl9rfUbV/wrHv7aJp0G3HA/rSWxUZ2ZRVhnO7pTnJ3Z9RmpfsaHGCV 5pktqd5Ac8cdKq2puqnuPUW1dhcYHcEVeVj5iHGOf5gj+tUILaQXC7WB5q6sM4ZcgffXv7ilZXM5 VZX3K/iKUR61dDcBlg2CfUA/1oqPxOjNrdwxTqsZ5Gf4ForLQv2jNvU4VivGkjVVS4Xzwq8KjZPm IPo2SPZgKpzxq8bA9T0zWjdHOg6cT1MvX6ouazVJLEk5+b+grSN27CaUNkZm7YSuO/ftSeZtYE9j 0pbv/XPUacyJn0rRaaB7STadyWTJO5BnJxk9++fypPssjjPTHFXIwBbx4A7fyFOc9aEujJrTd00S 6fAhtwpySj4rQiYwvG8YG8MVGPUggfqVqlpv3Zf9+r0X8J7iVef+BLUvQ5r3eo7VTHBf3qpjyzcM 0YHUq/zj9GA/Cs24glljV8LGoP3nbHH0rRl51PceT5MfJ/65rVW85t5s84Peou2aJKOpFa5gRkWV JM/Mf3YOD0xyPap2divKxn6oKq233pPoKsfwH6/0quVWNqU23YhkCGSN3AjIOMxr/SrZhZ9yiaMj afvjHNU7r/j3b6j+dXQcuAeRik4mUp67FrUY5z9jlEaN5lquSjg8hmz+mKyL4SJPHugk5HZc/wAq 2b0D7Dp3HaT/ANp1kXrusse12HPY1kpNNopKLs7EAlVTghhz3UimTOhc/MBz347Vbt5pcH94/U/x H2q9/rImL/Mc/wAXNU6jTNvZpwZj2TjzwQQR9a0x1UdTuGPzqlLFGH4RRx2FV4mZXUKxA3jofanz X6GEoW6lvxA+7VpjsPRB/wCOCiqWsyP/AGnP87dR39hRRZE6n//Z ------=_NextPart_000_0000_01C85294.643E2190--