WINTER, Circuit Judge:

This factually complex litigation arises out of a dispute over the disclosure of documents, representations, and warranties made by National Distillers and Chemical Corporation ("Distillers") in connection with the sale of its wholly-owned subsidiary, Elkhorn Re Insurance Company ("Elkhorn"), to Delta Holdings, Inc. ("Delta"). Following a bench trial before Judge Keenan, the district court held that Distillers violated federal securities law, committed common law fraud, and breached various express warranties. The district court awarded Delta $24.3 million in damages plus pre-judgment interest and ordered rescission of the entire transaction. We find as a matter of law that Distillers neither omitted to disclose material facts, made material misrepresentations, nor breached its warranties. We therefore reverse.

BACKGROUND

Distillers, now named Quantum Chemical Corporation, is a diversified company primarily engaged in the business of producing chemicals and liquefied petroleum gases. Elkhorn was originally established for the purpose of acquiring and developing operating insurance or reinsurance subsidiaries to insure casualty and property risks of Distillers. Sometime thereafter, Elkhorn began to reinsure risks underwritten by other companies. The principal factual and legal issues on this appeal relate to contemporaneous (with the acquisition) determinations of the adequacy of financial reserves set aside by Elkhorn to cover future claims. An understanding of these issues requires a lengthy description of the evidence at trial, beginning with an overview of the methodologies of estimating loss reserves in the reinsurance industry.

1. Loss Reserves and Reinsurance

Risk-pooling is a form of diversification that reduces the dispersion or volatility of losses and is the essence of insurance. Reinsurance is the pooling among secondary insurers of portions of risks previously underwritten by primary insurers. In typical reinsurance transactions, primary insurers first underwrite risks in exchange for premiums from the insureds. To spread the underwritten risks further, primary insurers transfer or "cede" a portion of their risks to reinsurers, who accept the risks in exchange for premiums from the ceding companies. Reinsurers, in turn, may cede portions of their risks to secondary reinsurers or "followers" in what are commonly referred to as retroactive cessions.

Reinsurance contracts typically fall into two categories. A "treaty" is an agreement under which a reinsurer accepts a percentage participation in all risks of a certain type or class underwritten by the primary insurer (or another reinsurer) during a specified period of time. A "facultative contract" is an agreement under which a reinsurer assumes specific risks instead of an entire class of risks.

Reinsurers assume many types of risk by treaty or facultative contract. These include death ( e.g., life insurance), property loss ( e.g., fire insurance), and liability to third parties for personal injury or property damage ( e.g., professional malpractice insurance). The underwriting of third-party liability, known as "casualty risks," leads to complex problems of financing and accounting because assumption of third-party liability risks involves substantial delays or "tails" in the discovery and reporting of claims. These delays, as lengthy as fifteen or twenty years with some policies, such as medical malpractice insurance, inevitably create considerable uncertainty as to the calculation of future claims and of the reserves that must be set aside to pay those claims. Such calculations are at the heart of the present dispute.

In preparing periodic financial statements, a reinsurer must treat amounts of earned premiums as current income and amounts of future claims as offsets to current income. These loss reserves often represent the largest liability item on a reinsurer's balance sheet, and particularly the balance sheet of a casualty risk reinsurer. Loss reserves must be established for known claims ("case reserves") as well as for incurred-but-not-reported claims ("IBNR reserves"). Case reserve estimates are less conjectural than IBNR reserves because case reserves are established immediately after a specific claim is reported. Case reserves are thus sums set aside to cover estimated losses based on reported claims. In contrast, IBNR reserves are sums set aside to cover losses for which claims have not been reported but must be estimated so the company can pay future claims. For that reason, reinsurers that underwrite casualty risks with long discovery or reporting delays often carry IBNR reserves that dwarf case reserves.

Under generally accepted accounting principles ("GAAP"), a reinsurer is obligated to make a reasonable estimate of IBNR liabilities. However, GAAP neither specifies a precise actuarial method nor requires that the reinsurer retain an independent actuary to prepare or review loss reserve estimates. Pertinent to the instant matter are three methods of estimating IBNR reserves: (1) the incurred loss development method; (2) the loss ratio method; and (3) the Bornhuetter-Ferguson method ("B-F Method"). Each of these methods is well known within the reinsurance industry.

The incurred loss development method projects future claims by using data from past claims experience. Judgment calls as to selection of pertinent data and its use are inherent in the incurred loss development method. The loss ratio method utilizes a flat percentage of loss for each dollar of premium. Under that method, the percentage may be applied to the reinsured risks as a whole or different percentages may be applied to particular categories of risk or treaties with other companies. The selection of the particular percentage(s) is also a judgment call(s) and based largely on the selector's view of future losses. Many of the judgment calls needed to implement the loss development or loss ratio methods rely upon historical data as to loss reporting patterns.

The B-F Method is a hybrid of the incurred loss and loss ratio methods. It divides expected underwriting losses for each year into two categories — expected unreported claims and expected losses based on reported claims. As an account year matures, estimates of unreported claims are replaced by reported claims, thereby improving the accuracy of the ultimate estimate. To apply the B-F Method, therefore, a reinsurer must consider two parameters — first, the initial expected loss ratio and, second, the expected reporting pattern for a particular account year. The initial-expected loss ratio is selected on the basis of a variety of factors such as the general performance of the industry, the reinsurer's own historical loss ratio, the break-even loss ratio, and a comparison of expected reported losses with actual reported losses in previous years. However, because the initial-expected loss ratio is used only to the extent that claims are unreported, the ratio's importance for a particular account diminishes over time. In recent account years, the initial-expected loss ratio represents the lion's share of the final liability estimate, whereas in older account years, the ratio has a diminished effect on the final estimate because increasingly larger portions of the losses incurred during those years resulted from claims that have already been reported.

The second parameter in B-F analysis is the percentage of total losses, past and future, reported to date. This percentage is estimated on the basis of historical reporting patterns — i.e., the same reporting patterns that can be used to make direct extrapolations under the incurred loss development method. Reliable historical data on loss reporting patterns is thus even more essential to use of the B-F method than it is to use of the loss development and loss ratio methods.

Among the methods of presenting historical loss reporting patterns are formatted data sheets known as "loss development triangles." Such triangles consist of a left-hand column of account dates ( i.e., years in which policies covered by the reinsurance treaty were underwritten); a column to the immediate right stating claims reported during the first year; and additional columns to the right stating cumulative reported claims several years into the "aging" of a particular account. So arranged, the data resemble a triangle because cumulative claims figures are available for several years with respect to the oldest accounts but for one less year with respect to accounts beginning in the succeeding year, and so on. A hypothetical loss development triangle (000's omitted), prepared in 1986 and reflecting data through December 31, 1985, might appear as follows: Fig. 1 Account Year 1 2 3 4 5

Similarly, given reported losses in Fig. 1 during the first year of 1981 accounts of $7 million and reported losses at the end of five years of $9.7 million, one might conclude, applying the incurred loss development method, that for every $7 million in first-year reported losses, $2.7 million should be set aside as IBNR reserves to cover losses anticipated during the subsequent four years. Or, for purposes of the B-F Method, one might estimate from Figs. 1 and 2 that a particular percentage of total losses will be incurred within a given number of years. All of the calculations described along with others may also be used to arrive at the percentage(s) to be used under the loss ratio method.

Judgments must inevitably be made in the use of these calculations. For example, if loss development ratios regularly rise from one year to the next, an average of those ratios would probably understate future losses. Selection of a development factor based on the latest ratio and the rate of annual increase rather than the average would seem more reliable.

It must be emphasized that no actuarial method is so accurate that it eliminates conjecture in the calculation of IBNR liabilities. Even case reserve decisions involving reported claims entail uncertainty as to the amount of final loss. IBNR reserves, however, are far more conjectural because they must be calculated without knowing even the number of claims. Overly conservative loss estimates are no answer. Overestimated reserves are harmful because reinsurance premiums are competitive and a competitive return on investment is necessary to attract investors. Methods that cause substantial excess reserves to be set aside may cause losses to a reinsurer for lack of underwriting or investment.

Finally, in the reinsurance industry history may be an imperfect guide to the future, particularly with regard to casualty risks. The incidence of claims may change, the costs of defense may increase, and inflation may lead to unexpectedly high losses per claim. Even the conservative B-F Method relies on assumptions as to future events and conditions, that, if wrong, will lead to substantial errors in the final estimate.

Consequently, regardless of the actuarial method used, the preparation of, and reliance upon a net worth calculation in a balance sheet for a casualty risk reinsurer is based in large part upon informed guesswork. One cannot, therefore, expect equivalent certainty in a balance sheet's statement of loss reserves and its statement of more determinable items, such as outstanding principal and interest on debt instruments. It is for that reason that GAAP neither specifies a precise method of estimating loss reserves nor even requires that an actuary prepare of review loss reserve estimates. Although this opinion entails extensive discussion of loss development triangles, GAAP does not require that they be used in determining appropriate loss reserves.

This extended discussion of loss reserves and the reinsurance industry is in part only a prelude to an explanation of a final detail regarding loss development triangles central to the instant dispute. Because such triangles are designed to assist in estimating the amount of unreported claims as of specific dates, the triangles must accurately incorporate the lag in the reporting of claims to reinsurers if unreported claims are to be estimated reliably. Underwriting claims should thus be tallied in the year in which the reinsurer actually learns of the claims.

To illustrate, if, by some chance, claims amounts in Fig. 1 were based on the date of the report of claims to ceding companies or brokers — e.g., some claims reported to the reinsurer in 1982 would be listed under 1981, when the ceding company or broker learned of them, and so on through each year — the numbers listed in Fig. 1 might appear as follows: Fig. 3 Account Year 1 2 3 4 5

A final word is necessary on the detection of the use in loss development triangles of dates of claims reports to ceding companies or brokers instead of dates of reports to reinsurers. An actuary using Fig. 3 on the assumption that the cumulative losses listed for each account year where based on dates of reports to reinsurers would be unable to detect an error in that assumption simply by analyzing Fig. 3. However, if a new triangle including data for 1986 were constructed, the error would become apparent. Most of the loss amounts for the latest year in Fig. 3 would be increased as some of the claims reported to the reinsurer in 1986 would be allocated to 1985, the year in which those claims were reported to the ceding company or broker. (This assumes that the date of report to the reinsurer is never more than a calendar year later than the date of the report to the ceding company or broker.) The new triangle might appear thusly: Fig. 5 Account Year 1 2 3 4 5 6

2. Elkhorn's Reinsurance Activities

In 1972, when Elkhorn, which was licensed in Kentucky and New York, began to broaden its business by reinsuring third-party risks, Robert Norton became its president. Norton joined Distillers as an accountant in 1946. He became an executive in 1949 and a corporate officer in 1963. Norton had no actuarial training or managerial experience in the reinsurance industry.

Elkhorn's third-party underwriting continued to expand until, by 1983, outside business represented the largest portion of Elkhorn's activities. A substantial portion of Elkhorn's outside or "assumed" business consisted of reinsuring casualty and ocean marine risks with long delays or "tails" in the reporting of losses. As a result, assessments of Elkhorn's net worth substantially depended upon projections of future claims liability. To calculate IBNR reserves, Elkhorn used the loss ratio method. It recorded sixty-five percent of earned premiums as IBNR reserves unless a ceding company recommended another IBNR reserve level with regard to a particular treaty, in which event Elkhorn followed the ceding company's recommendation. The loss ratio method — specifically, the sixty-five percent formula with a later modification by which incurred losses were retained in loss reserves — remained Elkhorn's method for determining IBNR reserves until its acquisition by Delta.

As early as 1981, however, Norton and other Elkhorn executives became concerned over the accuracy of their loss reserve estimates. In October 1981, Norton asked an outside actuarial firm, Tillinghast, Nelson Warren ("Tillinghast"), to study Elkhorn's loss reserves and to recommend a more sophisticated actuarial method. As part of Tillinghast's written project outline, Greg Leonard, a Tillinghast actuary, proposed that Tillinghast recommend an actuarial method and calculate an appropriate level of loss reserves. After reviewing Leonard's proposal, Norton and Ramsey Joslin, Elkhorn's chief financial officer, instructed Leonard to proceed with the study and recommendation but not to calculate a suggested level of loss reserves.

In February 1982, Tillinghast completed its study and delivered three bound sets of a two-volume report ("February Report") to Norton. Norton gave one copy of the Report to Elkhorn's controller, James McGurty. Norton testified that he gave another copy to the company's chief underwriter, Terry Brewer, but at trial Brewer could not recall whether he actually received a copy. Norton kept the third set for his own use, placing it in the credenza in his office. McGurty kept the Report in his files.

The February Report did not explicitly state that Elkhorn's loss reserves were deficient. However, its discussion and exploration of methodologies did suggest problems with Elkhorn's IBNR reserve estimates. Addressing the merits of various actuarial methodologies, the February Report: (i) observed that the incurred loss development method "can lead to erratic and unreliable projections" because "a small swing in early reporting results in a very large swing in ultimate projections"; (ii) cautioned that the loss ratio method, with which Elkhorn was calculating its loss reserves, "has the advantage of stability, but . . . ignores actual results as they emerge"; and (iii) recommended that in the future Elkhorn determine its IBNR reserves by the B-F Method, which it described in detail.

Among various appendices to the report were detailed worksheets from which Elkhorn's IBNR reserves could be calculated according to the B-F Method. These calculations were not completed. The worksheets were based on loss development triangles prepared manually from Elkhorn's accounting records by Tillinghast. There was evidence at trial that, if the calculations had been completed, they would have disclosed an IBNR loss reserve deficiency of approximately $10 million. The February Report also noted that, when data based on treaty categories became available through computerized bookkeeping, a refined B-F analysis based on such data would be even more informative than the use of the worksheets in the appendices. As an interim measure, while Elkhorn would be computerizing its bookkeeping, Tillinghast recommended increasing Elkhorn's assumed loss ratio from sixty-five percent to roughly eighty percent.

Norton testified that he never completed the calculations demonstrating a $10 million deficiency in IBNR reserves because that calculation would have become obsolete as soon as treaty-category data became available upon Elkhorn's planned conversion to computerized bookkeeping. However, based on McGurty's testimony that Norton had stated that the February Report estimated a $10 million deficiency, the district court disbelieved Norton's denial of such a calculation, a finding that is not clearly erroneous.

Norton requested that Tillinghast prepare another report on three cancelled reinsurance treaties (collectively, the "Barrett Treaties") that had not been included in the February Report. In April 1982, Tillinghast delivered this second report ("April Report"), which, unlike the earlier one, included all requisite calculations and explicitly stated that Elkhorn faced IBNR losses on the Barrett treaties of approximately $13.2 million. Elkhorn at that time was carrying on its books IBNR reserves of only $2.3 million for these treaties. The total deficiency in IBNR reserves estimated by completing the worksheets appended to the February Report and by the April Report was thus about $20 million. After consulting with Brewer about the deficiency revealed by the April Report, Norton purchased a $10 million loss transfer policy from the Continental Insurance Company (the "Continental Agreement") to cover the Barrett Treaties in exchange for a $5 million premium.

The seeds from which the present dispute germinated were planted in late 1982 when the computerization of Elkhorn's bookkeeping was completed. This computerization was based on software called STREAM which Elkhorn purchased from another reinsurer. (The computer was in Kentucky and used for Distillers' other businesses, Elkhorn being too small to have its own system). STREAM allowed treaty-category analysis as recommended by the February Report, and Norton generated treaty-category data from STREAM to construct loss development triangles as of December 31, 1982. Using these triangles, Norton estimated a deficiency far below $10 million. Although there is no evidence or district court finding that Norton realized it at the time, the triangles were, as he testified at trial, "all wrong."

The problem with Norton's triangles lay in STREAM. Whatever merits STREAM might have had as a system for maintaining and retrieving records or for analyzing data for other purposes, it had a serious deficiency so far as the construction of loss development triangles was concerned. Dates of the reports of claims to Elkhorn were in a STREAM file but were not separately retrievable. Dates of reports of claims to ceding companies or brokers were separately retrievable as report dates. The only report date retrievable by STREAM was thus not the date of a report of a claim to Elkhorn, "book date," but rather the date of a report of a claim to the ceding companies or brokers, in the lexicon of this litigation, "account date." STREAM data thus produced triangles like Fig. 3, supra, instead of like Fig. 1.

John Cascio, Elkhorn's assistant controller, understood that STREAM reported claims reports as of account dates but did not discuss this issue with Norton. Cascio had no knowledge of the effect such data had on loss development triangles. For Norton's part, he may have known (Norton denied knowing, Cascio "assumed" Norton knew) that account-date data was being used but, if he did, there is no evidence that he knew that it would impair the predictive value of the triangles.

3. Delta's Acquisition of Elkhorn

At the time of Delta's acquisition of Elkhorn, the reinsurance industry had been suffering a protracted slump attributable to excess underwriting capacity and widespread inflation. Many reinsurance companies, especially followers unable to dictate terms and premiums, were battered by stiff price competition and underwriting losses. Elkhorn was no exception. From 1979 to 1982, the company suffered a series of underwriting losses, posting modest overall profits only because investment income exceeded those losses. Consequently, by 1980, Norton and other senior executives at Distillers began to believe that Elkhorn's business was, as the district court put it, "going sour."

In April 1983, Distillers discontinued all new third-party underwriting and began to explore ways of liquidating or selling its reinsurance business. Early that month, Norton contacted Arthur Deters of American Risk Management, Inc. ("ARM"), the entity then responsible for Delta's day-to-day management and later responsible for managing Delta's operating subsidiaries. At a meeting with Deters on April 6, 1983, Norton and Joslin disclosed Distillers' decision to discontinue Elkhorn's third-party underwriting and asked if ARM could assist Elkhorn in one of three ways — (1) managing an orderly wind-down (or "run off") of Elkhorn's reinsurance treaties, (2) making private and discreet inquiries about possible buyers for Elkhorn, or (3) proposing to Delta that it purchase Elkhorn's third-party reinsurance business.

ARM chose to pursue the third option, and in early May representatives of ARM and Distillers met in New York where they agreed to explore the possibility of selling Elkhorn to Delta for the book value price of $18 million. In June, senior underwriters from ARM, Lawrence Bell and Bryan Murphy, and an actuary from Peat Marwick, Alan Kaufman, visited Elkhorn's offices on Delta's behalf, interviewed the Elkhorn's staff and reviewed various underwriting records. They were told about Elkhorn's problems with the Barrett Treaties and about the $10 million Continental Agreement. Norton showed his December 31, 1982 loss development triangles, based on STREAM data, to Kaufman and indicated that Tillinghast had educated him as to the B-F Method. Norton did not reveal the existence of either Tillinghast Report but rather stated that he had learned the B-F Method without having to pay for a study.

Bell reported to Delta that Elkhorn was rather disorganized, that errors were found in computerized data regarding two randomly selected treaties, and that "a thorough audit . . . on all major accounts" was necessary. His report concluded with the statement that "a thorough IBNR review must be made." Kaufman reported to ARM that he disagreed with the methodology by which Elkhorn was estimating loss reserves and concluded that, on a brief review of methodology and subject to several "unknowns," those reserves were deficient by some $5 million.

After further meetings, on July 19 Distillers and Delta reached an agreement in principle to sell the capital stock of Elkhorn for the book value price of $18 million. At the time, the book value shown on Elkhorn's June 30, 1983 balance sheet was $26,472,000 — a figure that included both Elkhorn's third-party reinsurance business, which Delta wished to buy, and its captive business, which Distillers intended to retain. Consequently, the parties agreed that Distillers would prepare a June 30 balance sheet segregating third-party and captive business specifically for the merger. These terms and conditions were incorporated in a letter of intent dated July 25, 1983.

For Delta, the two most important aspects of Elkhorn's financial health were the value of its bond portfolio and the adequacy of its loss reserves. Kaufman testified that, like most companies of comparable size, Elkhorn did not have an actuary. John Ryan, an ARM executive who represented that firm in the Elkhorn acquisition, testified that he also knew that Elkhorn lacked actuarial expertise.

As a result, Delta's acquisition was conditioned on receiving an opinion from an outside actuarial firm, Conning Co. ("Conning"), as to the adequacy of Elkhorn's loss reserves, and an opinion from an outside accounting firm, Peat Marwick, Mitchell Co. ("Peat Marwick"), as to the accuracy of Elkhorn's June 30, 1983 balance sheet, including of course Peat Marwick's view of the adequacy of loss reserves. Delta offered to allow Distillers to name an actuary to participate in the loss reserves examination, but Distillers stated that it was satisfied with Conning.

The Stock Purchase Agreement contained numerous protective warranties by Distillers, discussed in greater detail infra. In Section 4 of the Agreement, Distillers agreed to give Delta's actuarial and auditing representatives "reasonable access" to its books and records. In Section 5(f), Distillers warranted the completeness of its books and records, the fact that they had been maintained in accord with accepted insurance practices, and their accurate reflection of Elkhorn's financial status. In Section 5(g), Distillers warranted that the June 30, 1983 balance sheet was maintained in accord with GAAP and fairly presented Elkhorn's financial position. Elkhorn further guaranteed in Section 8(g) that all tax liabilities had been provided for and guaranteed that the market value of its bond portfolio would be no more than $1,635,000 below its book value as of August 31, 1983. Finally, in Section 12 it was also agreed that, within roughly two weeks after the acquisition, Delta would prepare a balance sheet for September 30, 1983, and Distillers would reimburse Delta for any difference between the net worth as shown on that balance sheet and $18 million. Any dispute over the balance sheet was subject to a binding decision by Peat Marwick.

Delta's representatives thereafter examined Elkhorn's books and records and Norton explained the business and actuarial practices of his company to Delta's representatives. In an August meeting with Ryan of ARM, Robert Brian, the actuary heading up Conning's study, and Gary Ransom, also of Conning, Norton explained that Elkhorn had been calculating its loss reserves either by applying the flat sixty-five percent loss ratio or by following the recommendation of a ceding company. In that conversation, he mentioned that he had tested Elkhorn's loss reserves by applying the B-F Method which, he said, he had learned from Tillinghast. According to Ryan, Norton said he obtained this instruction without having to pay for it and never mentioned the existence of either Tillinghast Report, although he was asked whether Elkhorn had had actuarial studies done. It is agreed that, at no time prior to the acquisition, were the Tillinghast Reports physically given to representatives of Delta.

On August 26, 1983, Conning delivered a written report to ARM concluding that Elkhorn had a loss reserve surplus of approximately $7.5 million as of December 31, 1982. Shortly thereafter, Conning revised its estimate and opined that Elkhorn's reserve surplus was about $1.6 million, adding the caveat that actual losses "may vary significantly from our estimates since underlying data is quite variable and difficult to project." Conning appears to have been using loss development triangles based on STREAM, and thus on account-date data.

During the same period in which Conning was preparing its assessment of Elkhorn's loss reserves, Peat Marwick's auditors spent some 327 hours examining Elkhorn's books and records. In the course of this effort, loss development triangles based on claims up to June 30, 1983 were developed from STREAM data. In mid-September, Amy Factor, a Peat Marwick actuary, noticed that some amounts of reported claims on the December 31, 1982 loss development triangles differed from the amounts of reported claims for the same time periods on the June 30, 1983 triangles, differences similar to the changes illustrated in Figs. 3 and 5, supra. Of course, the very fact of changes in amounts of reported claims for closed time periods revealed a problem, as described supra in connection with Figs. 3 and 5.

When Factor asked Norton why loss amounts for closed time periods were changing, he had no answer but referred her to other Elkhorn personnel. They in turn explained to Factor that the changes occurred because STREAM retrieved account rather than book dates for reported claims data. Factor informed either Kaufman or David Wasserman, another Peat Marwick actuary, of her discovery. At their instructions, she attempted to contact Conning but apparently never got through. Kaufman testified that he left a message at Conning for Brian detailing the facts concerning STREAM's use of account rather than book dates. Brian denied ever learning of this fact before Delta's acquisition of Elkhorn. Kaufman also testified that he informed Ryan of ARM about the Elkhorn triangles being based on account-date data. Ryan denied ever learning of this problem before the acquisition of Elkhorn.

In the glow of hindsight, the parties agree that calculating loss development triangles based on account-date data will, if not compensated for, result in a serious understatement of IBNR reserves. One of Distillers' own experts testified that proper corrections for the account-date distortion caused by Elkhorn's computer program might have revealed a loss reserve deficiency of as much as $108 million as of the date that Elkhorn's books represented a net worth of $18 million. The account-date distortion was, therefore, indisputably significant.

However, no substantial corrective action was taken as a result of Factor's discovery. Other than increasing loss reserves for the year 1982, Peat Marwick took no steps in response to the problem. (Brian of Conning and Ryan of ARM denied ever learning of it.) The failure of Conning to revise its prior reported opinion or to react to his phone message appears not to have troubled Kaufman. Kaufman did not perceive, or take steps to learn of, the peril in relying upon account-date data. There is no evidence of any effort to determine how the account-date data might be compensated for. No inquiry appears to have been made at the time of the possibility of altering STREAM to use book-date data. Moreover, STREAM output was based on raw data in Elkhorn's files and manual retrieval of book dates was obviously possible, as Tillinghast had done before Elkhorn computerized its bookkeeping in late 1982. However, Kaufman never sought, or even inquired about, manual assembly of book-date data. As noted above, had that data been obtained and incorporated into the loss development triangles, Elkhorn's insolvency would have been revealed. Nor, apparently, was consideration given to delaying the acquisition until book-date data was acquired.

One reason for the casual reaction to Factor's discovery appears to have been Kaufman's belief that the account-date problem affected only the loss amounts for the year 1982. He thus compensated for a possible underestimation of loss reserves only for that year. According to Kaufman's testimony, he and Wasserman conferred and "agreed that the loss ratios looked reasonable except for 1982 and the 1982 loss ratio out of the data did not look reasonable and we had adjusted it so we thought we had a reasonable conclusion." (This testimony speaks volumes about the degree of guesswork that goes into estimates of loss reserves). Of course, the account-date problem affected every year, as the discussion in connection with Figs. 3-5 explains. Although the issue does not affect our ruling, it is possible that Factor's discovery may have been based on numbers changing only in the year 1982. (The illustrations in her testimony concerned that year.) Kaufman may thus have assumed that claims reports for only that year were affected, missing the facts that account dates were built into prior years and reporting lags were thus understated throughout.

Kaufman testified that his belief that the account-date problem was limited to 1982 was based upon Factor or Wasserman having been so informed by Norton. This hearsay testimony — seemingly at odds with an auditor's responsibilities to carry out an independent investigation — was objected to and properly admitted solely to explain Kaufman's actions and not for its truth. Factor did not substantiate Kaufman's story and testified that Norton indicated that he did not understand the problem and referred her to Elkhorn personnel who accurately informed her as to what data was being used. Wasserman did not testify. (A memorandum by Factor states that Wasserman believed that use of account dates was "not a problem" and could be compensated for.) There is thus no competent evidence, nor did the district court find, that Norton or anyone else at Elkhorn misrepresented the nature or implications of the account-date data from STREAM on which the loss development triangles were based.

Thereafter, Peat Marwick certified Elkhorn's loss reserves. It concluded that Conning's estimates were somewhat optimistic and that Elkhorn would face an IBNR loss reserve shortfall of $3.5 million. In addition, Peat Marwick decided to adjust Elkhorn's bookkeeping on the Continental Agreement, thereby adding another $5 million deficiency to the $3.5 million IBNR deficiency. Based on these estimates, Peat Marwick advised ARM that it could not certify Elkhorn as fully reserved unless future liabilities were discounted to reflect the earning potential of Elkhorn's invested assets, a less conservative approach that, although arguably permissible under GAAP, had not been used by Elkhorn in the past. Only after Delta had agreed to discount future liabilities on the June 30, 1983 segregated balance sheet did Peat Marwick certify Elkhorn's reserves adding an explicit caveat that "[i]nsurers who establish claim reserves by applying selected loss ratios to earned premium (such as Elkhorn) often understate IBNR due to management's optimism in making loss ratio selections."

On September 30, 1983, Delta's acquisition of Elkhorn closed in Hamilton, Bermuda. Norton remained as president of the renamed entity, Delta Re; McGurty remained as controller. During the post-acquisition period, Peat Marwick reviewed the September 30 balance sheet pursuant to Section 12 of the Stock Purchase Agreement. Peat Marwick concluded that the balance sheet was in accord with GAAP and stated that it was unaware of any Appropriate modifications. Delta therefore did not request reimbursement on the ground that Elkhorn's net worth was less than $18 million on that date.

Although the inadequacy of account-date data was recognized by Delta and a project was undertaken to convert Delta Re's computerized bookkeeping to use of book dates, there was little urgency about the matter, and Delta Re continued to rely upon account-date data until the latter part of 1984, as Delta's own computers were gradually put into use. In fact, there is little evidence of any interest on the part of Delta's top management in the account-date problem until May 1984, when it was explored by Delta's advisory committee prior to a board of directors meeting. Even at this point, however, no one at Delta seems to have appreciated the full significance of the continued use of account-date data. In this period of time, Brian of Conning made further estimates of loss reserves based on loss development triangles containing STREAM data. Just as Kaufman thought that only the year 1982 was affected in Factor's triangles, Brian took corrective measures on his 1984 triangles only for the year 1983.

Meanwhile, Delta Re's fortunes declined further, as did those of the reinsurance industry generally, as a result of underestimated loss reserves. In July 1984, special examiners from the Kentucky Insurance Department began an investigation into Delta Re's financial condition, eventually concluding that the company's loss reserves had been deficient by some $38 million at the end of 1982. During the course of that investigation, state examiners told Norton that they planned to ask Tillinghast to perform a more detailed loss reserve analysis. Norton made no mention to anyone at Delta Re that Tillinghast had done work for Elkhorn in 1982.

On September 14, 1984, Norton resigned. John Ryan succeeded Norton as president, and, two months later, found the February Report in the credenza behind Norton's former desk. In early January 1985, Delta Re discovered the April Report elsewhere in Elkhorn's records.

On May 29, 1985, the State of Kentucky seized Delta Re's assets and commenced liquidation proceedings. According to state examiners, the company had been insolvent since the end of 1982. By the time that the company was seized, Delta had contributed some $6.3 million to its acquisition above and beyond the $18 million purchase price paid to Distillers.

4. The Proceedings in the District Court

On May 3, 1985, Delta commenced the instant litigation. The complaint alleged breach of various warranties contained in Sections 5 and 12 of the Stock Purchase Agreement ("SPA") (Count I); common law fraud and deceit (Count II); violations of Section 12(2) of the Securities Act, 15 U.S.C. § 77l (Count III); violations of Section 10(b) of the Securities Exchange Act, 15 U.S.C. § 78j, and Rule 10b-5 promulgated thereunder, 17 C.F.R. § 240.10b-5 (Count IV); a pattern of racketeering under the Racketeer Influenced and Corrupt Organizations Act ("RICO"), in violation of 18 U.S.C. § 1962(c) (Count V); violations of the New York General Business Law (Count VI); and negligent misrepresentation (Count VII).

On April 8, 1988, the district court granted summary judgment in favor of Distillers on Delta's claim for breach of Section 12 of the SPA. Observing that Section 12's net worth guarantee, see Note 4 infra, was accompanied by a host of procedures and remedies, including "final and conclusive and binding" arbitration before Peat Marwick. Judge Keenan held that Section 12 created no independent cause of action for breach of warranty beyond the procedures enumerated in the provision itself. The district court also granted summary judgment in Distillers' favor on Delta's racketeering claim, concluding that the misrepresentations, if any, were made in connection with a single acquisition and were insufficient to constitute the requisite pattern of "racketeering" acts under RICO.

The district court denied Distillers' motion for summary judgment on Delta's remaining claims, explaining that "[a]lthough Delta Holdings [through its agents, Peat Marwick and Conning] knew that Elkhorn used an account date basis to formulate its loss reserves, Delta Holdings was unaware of certain facts that could have altered its view of Elkhorn as an acquisition." Among these, suggested the court, were the "depth of Norton's knowledge" of the account-date problem and its effect on loss data from older account years; "whether Norton produced truly representative [broker] statements [from earlier account years], or whether certain statements were chosen in an effort to mollify [Peat Marwick's account-date] concerns"; and the circumstances surrounding "the non-disclosure of the Tillinghast documents." The district court also rejected Distillers' statute-of-limitations defense to Delta's claim under Section 12(2) of the Securities Act, reasoning that, under the adverse interest exception to the law of agency, Norton's knowledge may not have been attributable to Delta and that material questions of fact existed as to Delta's own knowledge.

Following a bench trial, the district court ruled in Delta's favor, based on (i) Norton's failure to disclose the Tillinghast Reports, which, in the court's view, would have disclosed Elkhorn's $20 million loss reserve deficiency and insolvency to Delta, and (ii) the court's conclusion that Distillers had guaranteed the accuracy of the June 30, 1983 and September 30, 1983 balance sheets' estimates of loss reserves. The district court found that Elkhorn was insolvent at those times, based on the "most credible and compelling explanations" given by Delta's expert witnesses of Elkhorn's conditions on those dates. It further found that Norton knew of Elkhorn's insolvency. This finding was based on Norton's request that Tillinghast not calculate actual loss reserve liabilities in the February Report and his concealment of both Tillinghast Reports. The court noted that use of account-date data rather than book-date data understated loss reserve liabilities. However, it found no deception in connection with use of the STREAM software.

The court held that Distillers, through its agents Norton and Joslin, violated Section 10(b) of the Securities Exchange Act and Section 12(2) of the Securities Act, committed common law fraud, breached Elkhorn's promise to give Delta access to Elkhorn's books and records, and breached its warranties of accuracy as to the June 30, 1983 and September 30, 1983 balance sheets. To restore Delta to its pre-purchase position, Judge Keenan rescinded the acquisition and awarded pre-judgment interest on the purchase price running from September 30, 1983. Moreover, because Delta's post-purchase capital contribution reasonably could have been anticipated at the time of the acquisition, and because Kentucky's seizure of the company had resulted in the complete loss of that $6.3 million contribution, the court awarded damages in the full amount of Delta's capital contribution and interest from May 29, 1985, the date Kentucky seized Delta Re's assets. This appeal followed.

DISCUSSION

We briefly summarize our holdings. The contents of the Tillinghast Reports were either not material in the context of this transaction or were disclosed. The February Report contained information on the B-F Method that was well known to actuaries, including those at Conning and Peat Marwick. The projections that might have been made from its worksheets were stale at the latest by December 1982. The substance of the April Report, if not its existence, was disclosed to ARM and Peat Marwick at the earliest opportunity.

We conclude that the district court's finding that Norton knew that Elkhorn was insolvent is clearly erroneous. Disclosure and scrutiny of the February Report might have alerted an actuary to the account-date problem. However, Peat Marwick knew of this problem before the acquisition, and Elkhorn was under no duty to make an independent study of the effect of the use of account dates. Finally, we hold that Distillers did not guarantee the adequacy of Elkhorn's loss reserves estimates.

1. The District Court's Conclusions Regarding the Tillinghast Reports

The district court's finding that Norton failed to disclose the February and April 1982 Tillinghast Reports to Distillers and knew of Elkhorn's insolvency was the basis for its conclusion that Delta violated Sections 10(b) and 12(2), committed common law fraud, and breached the agreement to provide access to Elkhorn's books and records. The district court believed the Tillinghast Reports demonstrated Elkhorn's $20 million loss reserves deficiency and insolvency and were concealed for that reason. As noted, the district court found that neither Delta nor its representatives were told of either Tillinghast Report and that this non-disclosure was accompanied by "affirmative misstatements by Elkhorn-Distillers representatives to [Delta's] people." These misrepresentations concern Norton's statements regarding his learning of the B-F Method from Tillinghast. Judge Keenan thus placed great weight on Kaufman's recounting of a meeting with Norton in which Norton casually joked about learning the B-F Method from Tillinghast "without having a study from them . . . [and] without having to pay for it."

The district court's finding that Norton knew of Elkhorn's insolvency at the time of the acquisition is clearly erroneous. This finding was based on Norton and Joslin's request that Tillinghast not calculate a precise loss reserve figure in the February Report and Norton's subsequent concealment of both Tillinghast Reports. The finding thus dates Norton's knowledge of insolvency as beginning in October 1981, when the February Report was commissioned with a request that a loss reserve figure not be calculated, and continuing for some two years until the Delta acquisition. It is not supported by the record.

Assuming Norton could "know" of Elkhorn's insolvency in 1981 — given the imprecision of loss reserve estimates, Norton's basic ignorance of actuarial methodology, and the paucity of evidence that Elkhorn was actually insolvent in October 1981 — such a finding assumes the existence of a highly sophisticated, even fantastic, plot that has no evidentiary basis in the record. Such a scheme would have to have begun almost a year before there was any discussion with third parties concerning a sale of Elkhorn. It also would have to have involved knowledge of the effect of the use of STREAM software data on loss development triangles.

Based on STREAM data, Norton constructed loss development triangles for December 1982 that showed a relatively negligible loss reserve deficiency. If Norton knew of Elkhorn's insolvency from October 1981 to September 1983, then he would have had to have known of the account-date problem and of the magnitude of its effect on his December 31, 1982 triangles. Such a plot would have had to a rely on the hope that a proposed purchaser learning of the account-date problem would not seek book-date data before going on with the acquisition. Moreover, to succeed, such a scheme would require the cooperation of others at Elkhorn, at least McGurty and Cascio, who would have had to join in the fraud on the purchaser, a firm that was about to become their employer. There is no evidence of such a highly implausible scheme.

The evidence is that STREAM was purchased by Elkhorn in order to computerize its bookkeeping. Based on this record, no one anywhere knew at the time that STREAM would not produce the most reliable data for loss development triangles or, until the events leading to this litigation, ever focused on that problem. As a general bookkeeping software, STREAM had many uses, and there is no evidence that Distillers' purchase of STREAM was anything but innocent. The record indicates that in 1982 reinsurers of Elkhorn's size generally had neither computerized bookkeeping nor an actuary. There is no evidence that Elkhorn's contemplation of a conversion to computerized bookkeeping, which began before delivery of the February Report, ever took the construction of loss development triangles, a novelty at Elkhorn, into account. No suspicion can thus attach to the acquisition of bookkeeping software that was not well-designed for preparing loss development triangles.

The district court did not, of course, find that any such plot existed. Nevertheless, absent such a scheme, Norton cannot be found even to have suspected Elkhorn's insolvency after he constructed the December 31, 1982 triangles. We therefore conclude that the finding as to his knowledge of insolvency at the time of the acquisition is clearly erroneous.

The fact that Norton did not know that Elkhorn was insolvent at the time of Delta's acquisition did not, of course, release him from a duty to disclose the Tillinghast Reports if they contained material information. Before addressing the materiality of the Tillinghast Reports, we note that the district court's findings concerning the concealment of the Tillinghast Reports are not clearly erroneous. However, because inferences regarding materiality may be drawn from concealment, we summarize the record concerning that concealment.

With regard to the February Report, all witnesses seem to agree that Norton indicated that he had learned of the B-F Method from Tillinghast. Kaufman's testimony that Norton indicated he had learned the B-F Method without having to pay for a study differs only in detail from Norton's story that Joslin ordered only a methodology study without paying for calculations that Elkhorn could do itself. Norton, of course, testified that he mentioned both Reports to Delta's representatives. Ransom, a Conning actuary, testified that, at a meeting in August 1983, Norton said, in Ryan's presence, that Elkhorn had gotten a methodology study from Tillinghast. (The district court rejected Norton's testimony but did not mention Ransom's.) Also, it is undisputed that Norton gave McGurty a copy of the February Report without restriction as to filing or distribution. McGurty testified that he kept the February Report in his files and would have shown it to anyone from Delta who asked for it before or after the acquisition. (The district court did not discuss the undisputed evidence concerning McGurty's copy.) Brewer may also have received a copy, although the record is unclear. Finally, Norton neither destroyed nor took the February Report with him when he left. Rather, he left it where it would inevitably be found by Delta personnel. Norton may not have disclosed the existence of the Tillinghast Reports, but the record does not suggest any strenuous efforts to conceal them.

With regard to the April Report, it is undisputed that Norton informed Bell and Kaufman in June 1983 of the loss reserve deficiency resulting from the Barrett Treaties and of his attempt to resolve that problem by the purchase of the $10 million loss transfer policy from Continental. Although the existence of the April Report was not mentioned, the substance of its contents was thus disclosed at the first opportunity.

We make these observations concerning the record only to note the frailty of any inference of materiality that might be drawn solely from the apparent concealment. In truth, apart from light they might have shed on the account-date problem, the Tillinghast Reports are red herrings.

The applicable legal standard regarding the materiality of omitted information is whether "there is a substantial likelihood that a reasonable shareholder would consider it important" or "a substantial likelihood that the disclosure . . . would have been viewed by the reasonable investor as having significantly altered the 'total mix' of information made available." TSC Industries, Inc. v. Northway, Inc., 426 U.S. 438, 449, 96 S.Ct. 2126, 2132, 48 L.Ed.2d 757 (1976). We note that the application of this standard in the instant matter concerns the estimate of Elkhorn's loss reserves and the value of the omitted information to Peat Marwick and Conning, firms with actuarial expertise hired to make an independent inquiry with regard to the adequacy of those loss reserves. We conclude that as a matter of law the Tillinghast Reports were not material at the pertinent times.

The February Report described the B-F Method in detail, recommended its use, and appended work papers that would assist in applying it to Elkhorn. However, the B-F Method was in the public domain, and it is undisputed that Conning and Peat Marwick were fully aware of it. To the extent the February Report described the B-F Method and contrasted it with the loss ratio method, it would have been no more informative to Conning and Peat Marwick than a discussion of the differences between cash and accrual methods of accounting.

Moreover, the precise calculations that might have been made from the worksheets appended to the February Report would not have been of interest sixteen months later in June 1983. By the very terms of the Report, the February calculations would be stale by December 1982. First, the February Report explicitly stated that treaty-category analysis was superior to the treaty-by-treaty analysis employed on the appended worksheets. Treaty-category analysis was available in late 1982. Second, the data available for constructing the December 31, 1982 triangles was more current by at least a year than the data used in the appended worksheets. An actuary coming upon the February Report would not have bothered to complete the appended worksheets but would simply have assumed that any relevant data contained in the appended worksheets would either be reflected in the December 31, 1982 and June 30, 1983 triangles or supplanted by treaty-category data. The fact that the later triangles contained account-date data merely underlines the fact that the account-date problem, not the lack of access to the February Report, caused the injury to Delta.

To put the matter another way, if the June 30, 1983 loss development triangles had been based on book data, no one could claim even marginal relevancy for the February Report. Peat Marwick and Conning would have made their respective actuarial judgments based on those triangles. If their opinions were negative, the acquisition would have been halted or proceeded at a lower price. If their opinions were favorable and thus too optimistic, the acquisition would have proceeded with the resultant loss to Delta. In such circumstances, however, no blame could have attached to non-disclosure of the February Report. The sole relevance of the February Report is thus in the light it might have accidently shed on the account-date problem.

The existence of the April Report was similarly immaterial. In applying the B-F Method to the Barrett Treaties, the Report concluded there was a deficiency of $10.9 million in loss reserves with respect to those treaties. However, the problem with the Barrett Treaties was revealed by Norton to Bell of ARM and Kaufman of Peat Marwick at their initial meeting in June 1983. Norton disclosed the purchase of the $10 million loss transfer policy from Continental as his attempt to remedy the deficiency resulting from the Barrett Treaties. The fact that a document contained the facts concerning the problem with the Barrett Treaties was not significant.

The district court appeared to take the view that Norton's opining as to the adequacy of the loss transfer policy as a means of redressing the deficiency resulting from the Barrett Treaties was a misrepresentation. Even assuming that Norton's opinion was material in the context of independent investigations by professionals with more expertise than he possessed, Delta was not injured. Peat Marwick deemed the Continental arrangement inadequate under GAAP and compensated for it in certifying the reserves. It can hardly be contended, therefore, that material facts concerning the April Report were withheld.

The only significance of the Tillinghast Reports would thus have consisted in whatever light they might have shed on the account-date problem. The February Report might have been significant to an actuary who completed the appended worksheets, if the pertinent loss reserves figures in the February Report could be compared with those in Factor's June 30, 1983 triangles. Whether these figures could be compared is unclear because the worksheets employed treaty-by-treaty analysis while Factor's triangles used treaty-category analysis. If they were comparable — a matter not settled on this record — such an actuary would have noticed the changing numbers — much as the differences between Norton's December 1982 triangles and Factor's June 1983 triangles led her to discover the account-date problem.

The testimony of both Kaufman and Mary Hennessey of Towers Perrin indicated that the materiality of the February Report lay in whatever aid it might have given in illuminating and overcoming the account-date problem. Similarly, at oral argument, counsel for Delta conceded, as he had to, that the Tillinghast Reports contained nothing new so far as the B-F Method was concerned but argued that the disclosure would have revealed the account-date problem. Others, such as Brian, said that actuary reports showing a loss reserve deficiency of $20 million would have been "of interest." That testimony, which assumed that the appended worksheets would have been completed, does not alter the fact that the only pertinent matter that would ultimately have been revealed was the account-date problem. (The section immediately following discusses the relevance of what the February Report would have disclosed concerning the use of account dates.)

However, the argument that the Tillinghast Reports were material because of the light they would have shed on the data on which the December and June triangles were based does not sustain the district court's conclusion regarding their materiality. The district court found that the Tillinghast Reports were material because they facially demonstrated Elkhorn's insolvency, not because they would have revealed the account-date problem. In fact, the court did not find a material misrepresentation or omission concerning the account-date problem.

To reiterate, putting aside the account-date issue, the relevant portions of the February Report were either in the public domain (the B-F Method), stale (1981 treaty-by-treaty figures), or known to Peat Marwick (the Barrett Treaties, Continental transaction). Except for what light might have been shed on the use of account dates, disclosure of these Reports would not have changed events. The district court's theory of the materiality of the Tillinghast Reports, therefore, cannot be sustained, and no liability exists under Section 10(b) of the '34 Act or Section 12(2) of the '33 Act based on that theory. Similarly, the district court's view that Distillers' failure to provide the Tillinghast Reports to Delta breached Distillers' warranty in the Stock Purchase Agreement to provide "reasonable access . . . to all of Elkhorn's . . . work papers, books and records . . . for purposes of review and inspection" and to "furnish (Delta) with all such reasonable information concerning Elkhorn's affairs as the buyer my request" cannot be sustained. Because the district court's theory of materiality is erroneous and we do not view the warranty as covering immaterial information, we hold that, putting aside the account-date issue, the warranty was not breached.

2. Distillers' Liability for the Account-Date Problem

Because of the centrality of the account-date problem to this appeal, we will assess the legal significance of Elkhorn's entire conduct — including but not limited to the non-disclosure of the Tillinghast Reports — concerning the use of account dates.

It is undisputed that Peat Marwick knew, before certifying Elkhorn's reserves, that the Elkhorn loss development triangles were based on account rather than book dates. Norton turned over his December 1982 triangles to ARM and Peat Marwick at the first opportunity in June 1983. It was Factor's comparison of the December triangles with June 30, 1983 triangles that revealed the erroneous nature of the data produced by STREAM. After Factor made that comparison, no one at Elkhorn sought to conceal the cause of the observed discrepancies. Because the use of account dates was disclosed, the only material information not revealed concerned the magnitude of the distortion that use of account dates caused in loss development triangles. However, no one at Elkhorn suggested that the magnitude of the error introduced by the use of account dates was of large, small or any particular dimension. Nor is there evidence that anyone at Elkhorn knew that use of account dates would distort loss development triangles, much less that they knew the direction or size of that distortion.

With regard to the February Report's relevance to the account-date problem, extended scrutiny of the differences between its worksheets and the triangles based on STREAM data might, if the numbers were comparable, have disclosed the magnitude of the distortion. However, that can also be said with regard to scrutiny of the differences between Norton's December 31, 1982 triangles and Factor's June 30, 1983 triangles. (The fact that both Norton and Factor were using STREAM data would not have prevented discovery of the magnitude of the distortion because that magnitude results from the built-in feature of account-date data, as discussed in connection with Figs. 3 and 5.) Finally, and really stretching, the February Report would have revealed that book-date data could be assembled manually. However, Peat Marwick knew that STREAM output was based on raw data in Elkhorn's files but showed no interest in manual retrieval after Factor's discovery of the account-date problem. The Tillinghast Reports thus would have added nothing material to the information about the account-date problem that Peat Marwick had by September 1983.

Nevertheless, given the importance of the account-date problem, we examine Elkhorn's entire conduct regarding that problem in light of relevant legal criteria. We turn first to the question of whether Distillers may be liable under Section 10(b) of the Securities Exchange Act. Liability under Section 10(b) requires a material misrepresentation and a showing of scienter. See Ernst Ernst v. Hochfelder, 425 U.S. 185, 96 S.Ct. 1375, 47 L.Ed.2d 668 (1976). That test has not been met. For reasons stated. Elkhorn's personnel were ignorant of the ramifications of the account-date problem and of the relevance of the Tillinghast Reports to that problem.

Liability under Section 12(2), however, is more easily established. Again, because the only material information not provided concerned the magnitude of error caused by the account-date problem, the pertinent questions are: (i) whether Distillers' failure to take any position on the magnitude of error was an omission of a fact necessary to make statements that were made not misleading, and (ii) whether, if so, Distillers carried its burden of showing that in the exercise of reasonable care it could not estimate the error's magnitude. We conclude that Distillers prevails as a matter of law on both questions.

Elkhorn personnel did not characterize the use of account dates as causing a distortion of any particular magnitude. We see no basis for concluding that this was a misleading omission. Elkhorn was known to lack actuarial sophistication, the very reason why Delta had insisted on favorable independent opinions from Conning and Peat Marwick as conditions of the acquisition. The silence of the non-actuaries at Elkhorn could not, therefore, have led a professional actuary to believe the problem was trivial.

Moreover, Elkhorn carried its burden of showing that it did not behave unreasonably in failing to probe the magnitude of the account-date problem when Factor inquired as to the reasons for the numbers changing in the loss development triangles. Delta's agents, including two firms with actuarial experience and knowledge far exceeding that of any personnel at Elkhorn, were conducting an independent inquiry into Elkhorn's financial status, with particular concern for the adequacy of its loss reserves. Elkhorn personnel had no reason to expect that their views would be welcome on a matter as to which they were far less knowledgeable than either Conning or Peat Marwick, and might reasonably assume that such questions were for Peat Marwick and Conning to resolve. Elkhorn therefore exercised reasonable care and is not liable under Section 12(2).

For similar reasons, we conclude that Distillers did not commit common law fraud. There was no material misrepresentation by Distillers concerning the account-date problem that was relied upon in any way by Delta. The injury to Delta was caused by its misunderstanding of the problem, which was in no way the result of Distillers' conduct.

Finally, any omission by Elkhorn regarding the magnitude of the account-date distortion did not violate its promise to provide reasonable access to its books and records. Such a promise cannot extend to matters of which a party is ignorant but which might indirectly be revealed by otherwise immaterial records.

3. The June 30, 1983 Balance Sheet: Breach of Warranty

In the Stock Purchase Agreement, Distillers made two representations pertinent to the instant appeal regarding the segregated balance sheet of June 30, 1983. In Section 5(f), Distillers warranted that

[t]he books and records of Elkhorn are complete in all material respects and have been maintained in accordance with good business and accepted insurance practices and accurately reflect the financial condition and results of the operation of Elkhorn

In Section 5(g), Distillers represented that

the June 30th Balance Sheet and related statement of income are complete and accurate in all material respects, were prepared in accordance with Generally Accepted Accounting Principles ("GAAP"), and the June 30th Balance Sheet presents fairly the financial position of Elkhorn as at that date.

The district court found Distillers in breach of both warranties because (1) "Elkhorn had incurred substantial loss reserve obligations as of September 30, 1983 which were not disclosed in full," and (2) "[t]he June 30, 1983 Balance Sheet showed a net worth of $18 million rather than Elkhorn's true condition[,] which was insolvency." "Based on information available as of June 30 and September 30, 1983," the court explained, two expert witnesses "demonstrated that the loss reserve liability figures in Elkhorn's balance sheets were understated and the net worth correspondingly overstated." The court further reasoned that, under New York law, a plaintiff suing for breach of a warranty need not prove express reliance such as a change of position in reliance on a misrepresentation. CBS, Inc. v. Ziff-Davis Pub. Co., 75 N.Y.2d 496, 554 N.Y.S.2d 449, 452, 553 N.E.2d 997, 1000 (1990). Instead, any reliance is satisfied by "the express warranty . . . being part of the bargain between the parties." Id. at 453, 553 N.E.2d at 1001. Relying on this principle, the district court held that Delta had purchased the promise "that warranties contained in the Stock Purchase Agreement were true," and rescinded the Elkhorn acquisition "[i]n view of Distillers' breaches of material portions of the Stock Purchase Agreement."

We do not disagree with the district court's finding, based on the testimony of Delta's expert witnesses, that Elkhorn was insolvent as of June 30, 1983 and that this insolvency was due to the underestimation of loss reserves. But see the qualifications stated in Note 2, supra. We also do not disagree that the underestimation could have been detected from information available as of that date. Indeed, as noted above, had book-date data been derived manually or STREAM converted to produce such data when the account-date problem was discovered, the insolvency would have been disclosed subject again to the qualifications stated in Note 2, supra.

However, we disagree with the district court's view that the warranties quoted constituted a guarantee by Distillers that loss reserve estimates on Elkhorn's books would prove in the future to be substantially accurate. Taking an overview of the deal, Delta refused to go forward without obtaining independent opinions from Conning and Peat Marwick as to, inter alia, the adequacy of Elkhorn's loss reserves as of June 30, 1983. Delta also bargained for and received a period of time after the acquisition to reexamine Elkhorn's net worth as of September 30, 1983 and to be entitled to whatever reimbursement Peat Marwick determined.

Delta knew that Elkhorn lacked actuarial expertise, and its initial inquiries revealed a need for a thorough audit and raised serious questions about both Elkhorn's methodology and conclusions as to the adequacy of its loss reserves. Delta's knowledge of Elkhorn's lack of expertise and insistence upon independent actuarial and accounting opinions and post-acquisition arbitration, in an acquisition where the price was set at book value, hardly suggests that Distillers had agreed to guarantee loss reserves. Such an agreement would have been so one-sided so as to be implausible — one in which Distillers could do no better than break even. As the district court viewed the contract, Delta would reap handsome profits if the loss reserves proved excessive. If book value decreased as a result of the underestimation of loss reserves, Delta's losses would be covered by Distillers. An agreement so fraught with downside risk to Distillers and no hope of gain seems unlikely in light of Elkhorn's actuarial ignorance, of the substantial uncertainty regarding estimates of loss reserves, and of Delta's insistence on favorable actuarial opinions as a condition of the acquisition.

Although the district court relied heavily upon the analysis of Elkhorn's financial condition at relevant dates by Mary Hennessey of Towers Perrin and John Creamer of Arthur Young Co., Delta never explains what Hennessey and Creamer did that Conning and Peat Marwick could not have done in the course of their July-October inspections of Elkhorn. Nor does the record. Delta thus asks us to construe this contract to allow recovery of the full purchase price based on professional expert testimony in court that disagrees with the contemporaneous professional opinions expressed by Conning and Peat Marwick, firms hired by Delta itself to render such opinions.

Turning to the precise terms of the Stock Purchase Agreement, we find no support for that extraordinary view of the Agreement. The language of Section 5(f) warrants the maintenance of Elkhorn's books and records according to accepted practices in the industry so as to reflect accurately its financial condition. There is nothing in that language suggesting that estimated items such as future losses are guaranteed as to future adequacy. What is warranted is that the loss reserves stated have been set aside. Section 5(g) on its face warrants only that the June 30, 1983 balance sheet was consistent with GAAP and does not guarantee the accuracy of estimates of future claims to any extent beyond that required by GAAP.

Moreover, the Agreement contained specific guarantees of balance sheet items that would have been superfluous if Sections 5(f) and (g) guaranteed the accuracy of all estimated items on the balance sheet. For example, Distillers explicitly warranted that "[a]ll material tax liabilities . . . are adequately provided for" and agreed to defend and satisfy liabilities in excess of Elkhorn's tax reserves. Distillers also certified that "the book value of the bond portfolio . . . did not, at August 31, 1983, exceed the market value thereof by more than $1,635,000," thereby minimizing the risk to Delta that Elkhorn's assets were inflated on the June 30 balance sheet. This warranty is significant because, in June 1983, ARM viewed the two greatest risks with regard to the acquisition of Elkhorn as the possible overvaluation of its bond portfolio and the possible inadequacy of its loss reserves, both of which are balance sheet items.

Finally, Section 12, which we set out in the margin, provided Delta with a postpurchase right to challenge the book value of Elkhorn. Under that provision, Delta was allowed to deliver no later than October 17, 1983 its own balance sheet setting forth Elkhorn's financial condition as of September 30, 1983. If Delta's balance sheet showed a net worth in excess of $18 million, Delta would pay Distillers a dividend equal to the excess amount; if it showed a net worth of less than $18 million, Distillers was required either (1) to remit to Delta the amount of the deficit by October 31, 1983 or (2) to submit the valuation dispute to Delta's outside accountant, Peat Marwick, for "final and conclusive and binding" resolution. Thus, Section 12 afforded Delta a binding, post-purchase opportunity to arbitrate book value, and thus the adequacy of Elkhorn's loss reserves, before Delta's chosen accountant.

We cannot, therefore, stretch the basic accounting warranties of Section 5 to serve as guarantees of the future adequacy of loss reserves without upsetting the allocation of risk deliberately established by the Stock Purchase Agreement. It is axiomatic that "[t]he term 'generally accepted accounting principles' should not be interpreted in vacuo but only in relation to the particular type of business involved." Pittsburgh Coke Chemical Co. v. Bollo, 560 F.2d 1089, 1092 (2d Cir. 1977). Likewise, the phrase "complete and accurate in all material respects" should not be divorced from the commercial context in which it is used. The presence of unequivocal warranties as to the adequacy of Elkhorn's tax reserves and the market value of its bond portfolio — items included in the June 30 balance sheet — strongly indicate that the parties viewed Section 5 as being no more than what its language says, a warranty as to accounting accuracy and regularity, not a blanket guarantee of net worth.

Under New York law, "[a] specific provision will not be set aside in favor of a catchall clause." William Higgins Sons, Inc. v. State of New York, 20 N.Y.2d 425, 284 N.Y.S.2d 697, 699, 231 N.E.2d 285, 286 (1967). "Definitive, particularized contract language takes precedence over expressions of intent that are general, summary, or preliminary." John Hancock Mutual Life Ins. Co. v. Caroline Power Light Co., 717 F.2d 664, 669 n. 8 (2d Cir. 1983) (applying New York law). If either Section 5 or Section 12 of the Stock Purchase Agreement is a guarantee of Elkhorn's net worth and creates rights of challenge to the June 30 balance sheet, Section 12 is obviously the more "specific provision" and Section 5, by comparison, the "catchall clause." William Higgins Sons, 284 N.Y.S.2d at 699, 231 N.E.2d at 286.

For these reasons, we believe Sections 5(f) and (g) warranted only that (i) no material item had been omitted; (ii) each item was accurately described, e.g., amount of loss reserve; and (iii) the June 30 balance sheet was prepared in accordance with GAAP. So read, Section 5 neither duplicates other warranties of specific balance sheet items nor undercuts the conditional guarantee and right of challenge provided by Section 12. So read, Section 5 affords Delta a right to challenge the propriety of Distillers' accounting under GAAP on the June 30 balance sheet, whereas the right to challenge loss reserve estimates otherwise consistent with GAAP is subject to the terms and remedies enunciated in Section 12.

Because the district court believed that Sections 5(f) and (g) guaranteed that the June 30 balance sheet statement of loss reserves would, with an excess of $18 million, be adequate to cover future claims, it never explained how — or even whether — Elkhorn's June 30 balance sheet ran afoul of GAAP. Although the district court found that "the actuarial analyses of Towers Perrin, as presented by Mary Hennessey, and the conclusions of Arthur Young Co., as presented by John Creamer, were the most credible and compelling explanations concerning Elkhorn's true financial picture as of June 30," Section 5 never warranted that the June 30 balance sheet represented the "most credible and compelling" portrayal of Elkhorn's "true financial condition." The district court's findings and conclusions thus have no relevance to the question of whether the June 30 balance sheet conformed with GAAP. We turn now to that question, the most easily resolved issue in this complex matter.

Delta does not claim, and the district court did not find, that the June 30 balance sheet was incomplete or inaccurate in any respect other than that its estimated loss reserves were too low. Peat Marwick certified that balance sheet as well as the September 30, 1983 balance sheet as consistent with GAAP. Delta's witnesses testified to no gaps or inaccuracies in the data base provided by Elkhorn's records, on which the June 30, 1983 balance sheet entries were based, apart from STREAM's failure to afford separate retrieval of book dates. However, Sections 5(f) and (g) do not warrant easy computerized preparation of loss development triangles, and no claim is made that the raw data in Elkhorn's files regarding book dates was inaccurate. GAAP does not require actuarial review of loss reserves estimates, much less the use of loss development triangles.

With regard to the estimate of loss reserves, Delta does not even argue that the actuarial method employed by Elkhorn — the loss ratio method — was inherently unreasonable or inconsistent with GAAP. As our extended discussion supra noted, GAAP does not require that reinsurers use any particular actuarial method, and the loss ratio method was recognized as acceptable.

Nor can it be maintained that the specific loss ratio employed by Elkhorn — sixty-five percent or the ceding company's recommendation — was at the time viewed by professionals as so overly optimistic as to violate GAAP. Delta presented no such evidence at trial and, had that been the case, Conning and Peat Marwick, both of which were fully aware of Elkhorn's methodology of estimating loss reserves, would have concluded their studies in less than an hour. As noted, loss reserves are not like a debt with fixed payments of principal and interest. Informed guesswork is an accepted basis for determining such reserves, and Elkhorn's books were based on such guesswork.

We therefore reverse.