Free Declaration - District Court of Delaware - Delaware

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 1 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 2 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 3 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 4 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 5 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 6 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 7 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 8 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 9 of 42

Exhibit A

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 10 of 42

IN THE UNITED STATES DISTRICT COURT FOR THE DISTRICT OF DELAWARE In re: Teleglobe Comm. et al., Debtors. ) ) ) ) ) ) ) ) ) ) ) ) ) ) Chapter 11 Jointly Administered Bankr. Case No. 02-11518 (MFW)

Teleglobe USA Inc. et al., Plaintiff, v. BCE Inc. et al., Defendants.

Civ. Action No. 04-1266 (SLR)

DECLARATION OF MARTIN FELSKY, Ph.D., J.D. DESCRIBING DEFENDANTS' E-DATA RETRIEVAL AND PRODUCTION PROCESS I, MARTIN FELSKY, declare as follows: Background 1. I am CEO of Commonwealth Legal, Canada's largest litigation support service bureau. Commonwealth Legal was incorporated in February 2000. We have six offices across Canada and employ over 65 full time staff, and over 100 contract coders. We have worked with virtually all of the top 30 law firms in Canada, as well as six of the largest law firms in the world. We have also worked with the Federal Department of Justice on major contracts and the provincial Attorneys General in Ontario and Alberta. 2. We are a full- service litigation support provider: we do consulting, software sales and support, and training, in addition to scanning, coding and electronic

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 11 of 42

discovery. We were instrumental in pioneering electronic appeal books in various courts across the country, including the Supreme Court of Canada. 3. We are the leading Summation software reseller in Canada (having won the Pinnacle Award for being in the top five resellers worldwide), the master IPRO Distributor (Platinum Partner and Silver Reseller), the exclusive Concordance and iCONECT dealers, and the leading Ringtail Alliance Partner. 4. As CEO of Commonwealth Legal, I advise major law firm and government clients in Canada, the United States and abroad on the deployment of litigation support document management tools and resources. Prior to founding Commonwealth Legal, I was Director of integer.actif, Canada's leading legal technology consulting firm. Integer.actif worked with more than 60 law firms and law departments to provide advice, training and support on strategic technology planning, large litigation cases, information systems implementation and training. Before founding integer.actif, I was Legal Technology Counsel and a research lawyer at McCarthy Tétrault, one of Canada's national law firms. 5. I obtained two doctorates from the University of Toronto (English and Law) and was called to the Ontario Bar in 1985. For the last three years I taught a course in Business Law and Ethics in the Executive MBA program at Concordia University. 6. I have written extensively on the subject of legal technology and am frequently quoted in Law Times, The Canadian Bar Association National, Canadian Lawyer and other legal publications. For almost fifteen years, I was the Chairman of the Legal Research Network. I am a member of both the Judges Technology Advisory

2

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 12 of 42

Committee of the Canadian Judicial Council and the Special Committee of the Task Force on the Discovery Process in Ontario. I have spoken at dozens of conferences throughout North America, in Europe and New Zealand, including the ABA TechShow in Chicago, and Legal Tech in Toronto, New York and Los Angeles. 7. Commonwealth Legal began working on electronic discovery matters in the Fall of 2002. We acquired one of the leading e-discovery family of applications, "Discover-E" from Pacific Legal and three of our staff were certified on the program. We are currently a Discover-E Premier Business Partner. In addition to Discover-E, we use HardCopy Pro PlusTM software from Extractiva (formerly Mobious Solutons Inc.). More information about this software is found at the Extractiva website: HardCopy Pro PlusTM is the most complete and robust professional-grade EDD solution, allowing service bureaus to provide rapid, enterprise-level electronic data discovery services to their clients. Easily make large batches of electronic documents and e-mails accessible to end-users for searching and review. Create images, capture document-specific metadata, deduplicate, bates stamp and export to your retrieval and indexing system. As the pioneer in this field, HardCopy Pro PlusTM has grown to be incredibly feature-rich. Litigation support service bureaus throughout the world have based their electronic data discovery services on our integrated solution. It's faster and more cost-effective than hiring programmers to write complicated and error-prone custom tools. HardCopy Pro PlusTM has flexible workflow, with years of EDD processing experience built-in. Mobious Solutions keeps up with industry trends and continually adds capabilities. We listen to the requests and feedback of the service bureaus we work with. It's no wonder that service bureaus use HardCopy Pro PlusTM to produce millions of pages each month. Trust your e-discovery to the solution of choice for large corporations and government agencies, including such key players as Pitney Bowes and the United States Department of Justice.

3

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 13 of 42

8.

We have processed millions of pages of documents for large law

firms and corporate clients in the manufacturing industry, financial services industry, and various other industries. Once data is processed, we often integrate the e-data with scanned images and provide clients with databases for their internal use, or a hosted webbased application for document review. Processing of BCE's E-Data 9. Starting in January 2005 and continuing through the months of

February and March, BCE shipped to Commonwealth Legal a total of some 230Gb of data stored on hard drive media and CDs, relating to the Teleglobe USA matter. 10. The 230 Gb consisted of: a. b. Hard drive data for certain custodians Server data that is identifiable for certain custodians (by surname or employee ID number appearing in the folder name) Unidentifiable server data ­ data in folders to which the custodians had access, but were shared by non-custodians.

c.

11.

We were requested to restore, index and search selected portions of

the data. In addition we extracted previously processed hard drive data for relevant custodians from a Canadian litigation database in order to expedite the review process. Approximately 85,000 records were extracted. In late January or early February 2005, Shearman & Sterling LLP requested that I help evaluate whether the parties' search terms would efficiently filter out clearly irrelevant documents. I had discussions with Diane Barrasso, Plaintiffs' IT consultant, about the effectiveness of using a long list of general

4

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 14 of 42

terms as a search strategy, and over a period of weeks we discussed various approaches to conducting the search of e-data in a more efficient way. 12. In February 2005, I ran searches using some of the search terms

proposed by the Plaintiffs on a sample collection of BCE documents containing 103,591 records. More than fifty percent of all the records in the collection were captured with only nine search terms: "value," "capital," "cash," "debt," "line," "finance," "Excel," "note," or "notes." That demonstrated to me that the proposed search terms were far too general. I sent an e- mail to Ms. Barrasso on February 22, 2005, which indicated that more than half of the documents were retrieved using only nine terms. In my e- mail, I also indicated the number of documents that were retrieved when certain search terms were used. 13. After restricting the searches to those documents in the collection

that were within the plaintiffs' proposed date range (January 2000 through December 2002), Commonwealth Legal concluded that almost 90% of the documents created during that time period were retrieved when we used the proposed search terms. Therefore, we concluded that the search terms did not effectively filter out irrelevant documents, and that they should be combined in order to have the relevant context. 14. In her response to my February 22, 2005 e-mail, Ms. Barrasso

thanked me for my "detailed" report. The next day, however, I received a new list of search terms from the Plaintiffs in which only one word had been eliminated ("Excel"), certain words had been added, no effort had been made to combine words, and the list contained many general terms, such as "line" and "note" which are far too common to be

5

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 15 of 42

meaningful without a combination. The Plaintiffs had essentially ignored my suggestion concerning the combination of terms. 15. I continued to have discussions with Ms. Barrasso in an effort to

achieve a more logical method of searching, but the Plaintiffs did not engage in a meaningful effort to combine terms. One of the most basic suggestions I made to limit the number of retrieved documents was to combine the listed search terms with "Teleglobe" and related terms. Even though the proposed search terms were still far too common (the list contained terms, such as "third," "share," or "note"), the use of the "Teleglobe" restriction would filter out some irrelevant documents. The Plaintiffs did not agree to my suggestion. They raised the possibility that this approach might be too narrow since in their view not every Teleglobe-related document necessarily contains the word "Teleglobe" (or a variation). In order to accommodate the plaintiffs, I suggested considering a search that would exclude non- Teleglobe related documents by using a two-stage search process: a. b. 16. Search 1 on all data in the date range = Teleglobe, etc. AND search terms Search 2 on the remaining data = Search terms BUT NOT "Project A" (where Project A is an unrelated project).

While I made that suggestion, I was concerned that Search 2 would

not necessarily be effective if (a) there were only a few non-Teleglobe projects or (b) there were many documents that contained references to Teleglobe and non-Telegloberelated projects. At that stage, our discussion was theoretical. It was therefore not proven that the use of non-Teleglobe related transactions would actually filter out many

6

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 16 of 42

documents. Nevertheless, I suggested that we run Search 2 in an attempt to move forward. 17. Even though no agreement was reached on the searches, Shearman

& Sterling LLP requested that Commonwealth Legal process data for certain custodians in an effort to produce their electronic documents prior to their depositions. Accordingly, we processed such data on an expedited basis and provided to Shearman & Sterling LLP the e-data of three deponents, Elizabeth Kavanaugh, Elie Daher, and Yanick DeGrandpre. It is my understanding that documents resulting from a search for "Teleglobe (or a variation) and search terms" were produced to the Plaintiffs prior to those depositions. 18. By mid-March it appeared that there was no agreement on the

appropriate search terms, and Commonwealth Legal was instructed by Shearman & Sterling LLP to process the e-data and create a complete repository of all the documents that could be extracted for the relevant custodians. The purpose of that approach was to ensure that, as the negotiation on search terms proceeded, BCE would be able to respond quickly to different scenarios. On March 20, 2005, we began full processing of the data. That is not the same as restoring and indexing. Processing includes the following steps: a. Extracting all native files from hard drive media that had not already been restored b. Loading all data into our Extractiva servers c. De-duplicating all data so exact duplicates are not processed more than once d. Extracting any zipped files or other "container" files into their constituents e. Converting Netscape e- mail to PST format

7

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 17 of 42

f. Extracting all metadata and e-mail header information into a database g. Extracting all full text from native files and convert to ASCII format h. Converting all files to TIFF image format 19. In a letter dated March 30, 2005, BCE reiterated that using the

Plaintiffs' proposed search terms with "or" connectors would result in retrieving approximately 90% of the documents created in the period of January 2000 to December 2002, based on a sample of easily accessible data. BCE also indicated in the letter that it had regularly updated the Plaintiffs on the progress of its evaluation regarding the documents that would be retrieved using those search terms, but that Plaintiffs had not provided reciprocal disclosure. BCE stated its belief that too much time had been spent in this process, and that it intended to conduct Search 1 and Search 2. BCE attached to its March 30 letter a list of 147 search terms. In light of Plaintiffs' reluctance to combine search terms, the list attached to BCE's March 30 letter only reflected an effort to combine the most obvious terms. For instance, I understand that one of the issues in this case involves Third Series Preferred shares. Plaintiffs had sought the review of all documents that contained the word "third." In its March 30 letter, BCE indicated that it would not agree to search for the individual word "third," but for the combination "third series preferred." 20. I was told that, on the very next day, the Court held a hearing to

address, among other things, the search of e-documents. At the March 31 hearing, the Court indicated that it had always assumed the parties would use phrases with connectors, not individual words, in conducting searches in order to put words in context. That was precisely the approach that I had recommended over a period of several weeks. After

8

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 18 of 42

learning about the Court's remarks, I was hopeful that the parties would be able to agree on a more logical method of searching for electronic documents. 21. The parties convened a "meet and confer" on April 6 in New York,

which I attended. At that meeting I was informed by the Plaintiffs that they had finally agreed to combine lists of search terms with "Teleglobe" and related names. I thought that was an important breakthrough. The Plaintiffs also restricted some of the terms on their original list by combining terms with Boolean "and" "or" proximity connectors, but they refused to consider any phrases other than the few, most obvious, combination of words that BCE had suggested before hearing from the Court and a few additional modifications that Plaintiffs made at the meeting. As a result, BCE would still have to review in Search 1 all documents that contained "Teleglobe" (or a variation) AND one or more of 147 search terms, most of which were quite general, such as "share," "shares," "line," "dilut*," "projected," "value," or "treatment." In an effort to move the process along, BCE stated that it would run Search 1 on the hard drives of the custodians and the identifiable portions of their server folders. With respect to the contents of the unidentifiable portion of the server folders, BCE stated that it would conduct the following search: Surname or employee ID number and Search 1. The Plaintiffs also requested that a search be conducted involving Mr. DeGrandpre, which BCE agreed to conduct. 22. At the April 6 meeting, however, BCE indicated that, with respect

to Search 2, it could not agree to ignore the Court's views that the search should be conducted by phrases rather than words. BCE's position, which it subsequently reiterated in a letter dated April 15, was that once Search 1 was conducted (and the volume of 9

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 19 of 42

documents that is not retrieved in Search 1 was known), Commonwealth Legal would run the "negative" search to determine the extent to which that search functions as an effective filter. If Search 2 did not effectively filter out documents, the parties would have to search phrases as the Court had suggested at the hearing. 23. At the April 6 meeting, the Plaintiffs also asked for lists of the

following folders located on hard drives and serve rs, which BCE agreed to produce: (i) a list of folders generated from a search of the identifiable data of the custodians, using the following folder names: Teleglobe, TGO, TCC, THC, THUS, THUSC, TI, TINC, TTC, TUSA, Globesystem, Globe System, and TCLP (86 folders); (ii) a list of folders located on the unidentifiable areas of the servers (i.e., folders that cannot be readily attributed to a particular custodian), but to which one or more of the custodians had access and that contain in their folder names one of the thirteen "Teleglobe" terms mentioned above (721 folders); (iii) a list of folders, which consists of all folders on the BCE network to which one or more of the custodians had access (30,901 folders); and (iv) a list of identifiable folders, which represent all identifiable folders of the 37 custodians (2,063 folders). 1 The Plaintiffs asked whether BCE could indicate the amount of data contained in each folder. BCE undertook to provide that information only if it could be generated easily and automatically. Commonwealth Legal subsequently determined that it could not be. 24. On April 12, 2005, Plaintiffs' counsel wrote to BCE's counsel to

submit a revised version of the list of search terms reflecting Plaintiffs' proposed changes
1

The lists of 86 and 721 folders were provided to the plaintiffs on April 15, 2005. The lists of 30,901 and 2,063 folders were provided to the Plaintiffs on April 20, 2005. A final list of custodian hard drive folders was provided to the Plaintiffs on May 7, 2005.

10

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 20 of 42

on April 6 (attached hereto at Exhibit A) and to make additional requests, such as a request that BCE provide lists of additional folders. Commonwealth Legal agreed to the following deliverables: a. Re-do the folder search in the IDENTIFIABLE data (hard drive and server) by adding the strings "TCLP" and "globesystem," which Plaintiffs requested on April 12. If the search string "globe system" worked, include it. Provide a Concordance database of all the documents in those folders (which amounted to 86 folders). b. Re-do the folder search in the UNIDENTIFIABLE data (hard drive and server) by adding the strings "TCLP" and "globesystem." If the search string "globe system" worked, include it. c. In the IDENTIFIABLE data WITHIN the date range (excluding all records in the 86 folders previously provided), perform Search 1. Provide the number of responsive records. d. In the IDENTIFIABLE data WITHIN the date range that is NOT responsive to Search 1, and excluding all records in the 86 folders previously provided, search using the search terms in the April 12 letter but EXCLUDE documents containing the non-Teleglobe project terms and provide the resulting numbers to indicate how many documents have been excluded. e. In the UNIDENTIFIABLE data WITHIN the date range, use Search 1 AND Custodian (i.e. custodian surnames, employee numbers, assistants' surnames and assistants' employee numbers IN the document full text and the extracted metadata) and provide the resulting numbers. f. In the UNIDENTIFIABLE data WITHIN the date range that is NOT responsive to "Search 1 AND Custodian", search using the search terms in the April 12 letter and provide the resulting numbers. In the UNIDENTIFIABLE data WITHIN the date range that is NOT responsive to "Search 1 AND Custodian," search using one of the search terms in the April 12 letter but EXCLUDE documents

g.

11

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 21 of 42

containing the non-Teleglobe project terms and provide the resulting numbers to indicate how many documents have been excluded. h. In the entire repository unrestricted by any other terms, search WITHIN the date range for "degrandpre or de grandpre or degrandpré or de grandpré and commit*" and provide the resulting numbers. 25. The processing of data by Commonwealth Legal was conducted

simultaneously on five servers all dedicated to the project. The servers were running 24 hours a day, 7 days a week. The automated computer processes involved took some time, as they generated various dialog boxes, queries and anomalies that had to be manually attended to by our e-discovery specialists. After the processing was complete, various steps were required to create a searchable database, including Bates numbering, restructuring of file and folder contents, indexing of the data, and loading into Concordance. 26. By April 24, we had a searchable database of all the potentially

responsive documents. The complete repository contains approximately 6,000,000 pages. 86 folders of Identifiable Data 27. On April 15, 2005, Commonwealth Legal provided to Shearman &

Sterling LLP a database containing the docume nts located in the 86 folders that were created by one of the 37 custodians and contained "Teleglobe" or a variation in their title. There were 16,715 documents in those folders (counting each parent document and each attachment as separate documents), representing approximately 87,248 pages. It is my understanding that Shearman & Sterling LLP has offered to review all the documents

12

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 22 of 42

contained in those folders without resorting to an automated search, and that the responsive non-privileged documents will be produced by May 16. Search 1 28. We conducted Search 1 (i.e. Teleglobe-related terms AND

Plaintiffs' revised search terms) on the Identifiable Data within the date range. This search yielded approximately 26,468 documents. We created a database of those documents (excluding all records in the 86 folders previously provided to Shearman & Sterling LLP) and provided them to Shearman & Sterling LLP on April 25, 2005. 29. We also conducted "Search 1 AND Custodian" in the

Unidentifiable Data within the date range. That search yielded approximately 7,994 documents. We created a database of these documents and provided them to Shearman & Sterling LLP on May 5, 2005. It is my understanding that the documents retrieved in Search 1 (both the identifiable and unidentifiable data) are currently being reviewed by Shearman & Sterling LLP and will be produced to Plaintiffs by May 16. DeGrandpre search 30. On May 5, 2005, Commonwealth Legal provided to Shearman &

Sterling LLP a database containing the 121 documents resulting from the search Plaintiffs had requested. It is my understanding that the documents are currently being reviewed by Shearman & Sterling LLP and will be produced to Plaintiffs by May 16.

13

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 23 of 42

Search 2 31. We conducted Search 2 in the universe of documents not captured

by Search 1. Specifically, we conducted Search 2 in the Identifiable Data that was not responsive to Search 1, and excluding all records in the 86 folders previously provided. In conducting Search 2, we used search terms that denoted irrelevant transactions in an attempt to filter out irrelevant documents. 2 That search yielded a total of 51,562 documents created in the period January 1, 2000 through December 31, 2002, representing approximately 350,000 pages of documents. 32. We then conducted Search 2 in the Unidentifiable Data within that

date range that was not responsive to "Search 1." That search initially yielded 77,763 documents. At the April 25, 2005 hearing, Plaintiffs' counsel suggested that Search 2 in the Unidentifiable Data should be restricted by limiting the results to documents that contained custodian information in the document itself (i.e., surname or employee ID number). That revised Search 2 generated 13,620 documents. In the last few days, Commonwealth Legal obtained from BCE a more complete list of employee ID numbers and surnames for the custodians' assistants, and we determined that the number of documents generated by using those ID numbers and surnames is 14,035, amounting to approximately 98,000 pages of documents. As more employee ID numbers and assistant surnames and numbers become available, further searches will be performed and the results provided to Shearman & Sterling LLP.

2

Attached hereto as Exhibit B is a list of non-Teleglobe related transactions that was used in conducting Search 2.

14

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 24 of 42

33.

Thus, the total number of responsive documents generated after

conducting Search 2 on the Identifiable and Unidentifiable Data is approximately 65,597 (approximately 460,000 pages of documents), which amounts to almost twice the number of documents retrieved in Search 1. We were asked by Shearman & Sterling LLP to provide a random sample of 25 records from the Search 2 results and these were provided on May 5, 2005. It is my understanding that none of the documents in this sample related to Teleglobe and that they consisted mainly of non-responsive board minutes and corporate materials of non- Teleglobe entities. 34. In March 2005, I had suggested that one way of trying to improve

the precision of Search 2 would be to exclude documents relating to irrelevant projects. Once we were able to actually perform that search, it was clear that the strategy was somewhat successful in focusing the results to a more relevant collection, insofar as 48,562 unidentifiable and 30,045 identifiable documents were excluded (total 78,607). However, the disproportionately large amount of remaining documents generated in Search 2 demonstrates that an additional, or alternative, narrowing search strategy is needed to bring the collection to a more manageable and more responsive size. Alternative Search 2 35. One method to further cull the documents captured in Search 2 is

to use phrases, as the Court had suggested at the March 31 hearing, in order to put words in the proper context. While it is very likely that most of the responsive documents will refer to "Teleglobe," to the extent those documents might relate to Teleglobe without specifically naming it, I believe that a more logical way of capturing such responsive

15

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 25 of 42

documents would be to put the search terms in relevant context by using phrases. Attached as Exhibit C is a list of search phrases that have been proposed by BCE to further narrow the results of Search 2. Those search phrases were formulated from, and track closely, the language used in the Debtors' document requests. It is my understanding that a list of these proposed search phrases has been provided to the Plaintiffs for their review. Used as an alternative to the "negative" search, the proposed phrases capture 40,925 documents out of a total of 326,502 documents not captured in Search 1. That number of documents is significantly lower than the 65,597 documents captured using Plaintiffs' proposed Search 2, demonstrating that the use of phrases may be more effective in targeting responsive documents. Refined Search 2 36. I believe that an even more effective approach would be to use the

proposed phrases in addition to Search 2, namely, by applying the phrases to the 65,597 documents retrieved in Search 2. Such a refinement of Search 2 through the use of phrases ensures that the search terms are placed in relevant context, and therefore serves as a logical tool for filtering out irrelevant documents. 37. When the proposed phrases are applied to the 65,597 documents

retrieved in Search 2, the total number of documents captured (exclusive of the parents and attachments that were not responsive to the search terms) is 19,933. Thus, by refining Search 2 in this manner, the number of responsive documents is reduced by approximately 70%. I believe that this refined Search 2 method is significantly more effective at targeting potentially responsive documents than merely conducting Search 2

16

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 26 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 27 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 28 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 29 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 30 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 31 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 32 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 33 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 34 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 35 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 36 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 37 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 38 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 39 of 42

Exhibit B

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 40 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 41 of 42

Case 1:04-cv-01266-SLR

Document 145

Filed 08/01/2005

Page 42 of 42