Claimant seeks reimbursement for sums it claims it expended in performing a contract with the Illinois Office of Education (IOE) to provide meals to school children pursuant to the National School Lunch Act. 42 U.S.C. sec. 1751 etseq.
Claimant was a service organization or sponsor during the summer of 1977 under that Act. Claimant claims that it is entitled to reimbursement for meals provided in the amount of $541,530.00 but was paid only $360,491.00, leaving a balance due of $181,039.00.
The program is administered by the United States Department of Agriculture (USDA). Under this program the USDA grants money to the states to contract with service organizations or sponsors, such as Claimant, to *121provide free food services to children in poverty areas during the summer months. Claimant was allowed to and did contract with a food service management company, Open Kitchens, Inc., to actually prepare and deliver the meals to the sites. Federal regulations pertaining to the program are found in 7 C.F.R. sec. 225. The State of Illinois did not contribute any funds toward this program.
In 1977 the IOE agreed with the USDA to participate in the program. The IOE sent applications, handbooks and site handbooks to potential sponsors. The handbooks set forth the rules and regulations regarding the program (but did not set forth that statistical sampling would be used to audit payments under the program).
On May 10 or May 11, 1977, a training workshop was held at which time the program and the regulations were explained. On June 9, 1977, Claimant filed an application for sponsorship in the summer food program. Project sites such as schools, churches and community centers with summer children’s programs were selected and approved by the IOE. The children were to be served meals at the site. The Claimant had 167 sites. On June 24, 1977, Claimant signed a contract with the IOE (Claimant’s exhibit No. 1) which provided that Claimant would be a sponsor under the program and would receive reimbursement for meals served, and on June 27, 1977, Claimant commenced delivery of breakfast, lunch and supplement meals to the children at the sites. As of July 31, 1977, because a sampling monitoring program indicated a large number of ineligible meals were being claimed, the IOE cancelled the breakfast portion of the program and Claimant voluntarily cancelled the supplement meal portion of the program. The lunch program continued through August 27,1977.
Claimant presented a claim for reimbursement total-*122ling $541,530.00. According to Claimant’s certified public accountant, Sanford Aronin, of the C.P.A. firm of Checker, Simon and Rosner, the claim for reimbursement was done in accordance with the standards for audit issued by the Comptroller General of the United States and the United States Department of Agriculture audit guide No. 8270.6, dated March 1977. He further testified that Claimant had received the sum of $360,491.00 leaving a balance due of $181,039.00.
Respondent’s defense is that, based on an audit done on the program by the USD A, which audit was done on the basis of statistical sampling, the total amount reimbursable to Claimant was only $399,079.00, and that in addition to the $360,491.00 paid, there was paid, after the audit report, the sum of $38,588.00, thus leaving the amount at issue at $160,711.00.
On the question of the amount paid by Respondent to Claimant, the record is sparse. Claimant’s C.P.A. stated the amount paid was $360,491.00. The order of the U.S. Court of Claims, case No. 369-78, wherein the Claimant sued the United States directly and which case was denied on motion for summary judgment, indicated that only $360,419.00 was paid. The only evidence of payment of the disputed $38,588.00 was the testimony of Russell J. Hild, a witness for Respondent who was a supervisory auditor in the United States Office of Audit who testified as follows:
“Q. Do you have knowledge as to what actual amount has been in fact paid to Akiba for the 1977 summer lunch program?
A. It was up to what we had cited in the report, $399,079.00. And that wasn’t paid — I think at that time it was $360,000.00. But since then its been paid up to the maximum figure we had cited.
Q. So the total amount of the report, to your knowledge, has in fact been paid to Akiba pursuant to the figure of the audit report, is that correct?
A. Right.”
*123No date of payment was described. No cancelled checks, receipts or other documentary proof was offered as proof of payment. Claimant’s witness was an employee of the United States government and not of the State of Illinois and the source of his knowledge of payment was not brought out. Since payment is an affirmative defense, it is this Court’s opinion that Respondent has not sustained this burden as to the $38,588.00 and that there is therefore owed by Respondent to Claimant the sum of $38,588.00, regardless of the validity of the balance of the claim and of the Respondent’s defense thereto.
The bulk of Respondent’s defense is based on the USDA audit report which was done by statistical sampling and projections therefrom. The undisputed portion of the audit report gave 100% credit to Claimant for meals claimed in sites not included in the sampling. The issue revolves around the audit report giving credit for eligible meals based on a statistical sampling of sites. Meals that were deemed ineligible were deemed so for various reasons among which were that the meals were short on delivery, in excess of needs, eaten off the feeding site, had spoiled ingredients, had missing components, and were not served within the required time frame.
An understanding of the operation of the audit and of the summer lunch program is therefore necessary.
Meals under the program are delivered to the sites. The supervisors of the sites are employees of the church or organization which runs the youth program at that site. They are not Claimant’s employees. Meals are, according to the testimony of Andrew Lee of Open Kitchens, Inc., not delivered or served in a controlled environment. The sites are located in ghetto areas and are sometimes physically chaotic. The children are young *124and often undisciplined. The sites ranged from a minimum of 30 children per site to 400 per site. The sponsors were not required to have an employee present at each site on each day.
By far, the most meals deemed ineligible were meals claimed at sites in excess of the number of children in attendance, Since attendance of the children at a site is voluntary, a sponsor cannot predict with any accuracy how many children will attend on any particular day. Whether every meal is eaten at the site is also sometimes difficult for the sponsor to control or even observe at times.
The USD A had monitors who made observations on some days on some sites. Their observations were recorded and the results imparted to Claimant via “fast reports”, thereby giving Claimant some opportunity to make adjustments in the number of meals ordered for each site and to improve its performance in the other ways necessary to rectify errors. There is evidence in the record that Claimant did, in fact, in some cases rectify errors and adjust meals ordered, but in other cases, Claimant failed to make appropriate adjustments.
The monitors were college students on summer jobs. They were given some minimal training as to what to observe and how to record their observations. However, counting large numbers of meals to children who are not stationary could be difficult. All parties, including a monitor, Steve Lerner, agreed that some of the counting was a matter of judgment.
Normally, there were two monitors to a site although at times there was only one. Any differences in count as between the two monitors were resolved in favor of Claimant. The site supervisor, who was not a direct employee of Claimant, signed the monitors’ forms.
*125Larry Hartón was the auditor in charge and he reviewed the monitors’ forms. He testified that where there was a problem based on judgment he would resolve it in favor of Claimant. However, the most significant problem he noted was that Claimant continually made claims for leftover meals — meals delivered in excess of children receiving. However, cross-examination of Hartón and Lerner brought out several unexplained inconsistencies and potential errors.
According to the audit, approximately 48% of the total meals reported for the audited sites were counted as ineligible for reimbursement.
The monitors observed a total of 117 site-days (1 day’s operation at a feeding site) out of a total universe of 3,995 site days and projected the results at a 95% confidence level.
Dr. Herbert Arken testified for Respondent as an expert in the field of statistical sampling for audit purposes. He is an author of statistical auditing texts and numerous articles on the subject. He developed the statistical sampling plan which was ultimately used by the U.S. office of audit for the summer food program in 1977 and has been receiving $20,000.00 per year from the United States government for his expertise for many years.
Dr. Arken testified that probability sampling is widely used as an audit technique and is approved by the American Institute of Certified Public Accounts which sets the standards for the field of auditing and is used frequently by the Federal government and the Internal Revenue Service. There are a variety of sample plans available. The choice of a plan has a substantial effect on the magnitude of the sampling error. Simply stated, the *126more you sample, the less potential error there is in the result. If the sample is 100% there would, of course, be no error. In the case of this Claimant, the size of the sample and the plan used resulted in a 95% upper confidence level. This means that the audit is 95% certain that had the monitors examined every site on every day the value obtained would have been between the limits established from the sampling. In this case, the probability that the projection was incorrect was one in 20 or 5%. In the sampling plan, sampling error was taken into consideration so that, if the result were a mere projection of the samples, the amount found to be eligible would have been $44,000.00 less than that which finally resulted. To state it another way, allowance was made to Claimant for sampling error, in the amount of $44,000.00 because of the size of the sample even though it is 95% certain that the sampling error is something less than $44,000.00.
Dr. Arken testified that the audit calculations were correct and were done in accordance with generally accepted accounting practice and that an upper confidence level of 95% was the most commonly used figure in auditing.
On cross-examination, Dr. Arken admitted that errors in counting would alter the final result and that the effectiveness of his formula was dependent on the accuracy of the data which was gathered by the monitors.
Claimant called Dr. Haskell Benishay in rebuttal. Dr. Benishay is a professor at Northwestern University where he teaches economics, management, and statistics. Dr. Benishay testified that the statistical sampling audit did not take into consideration errors in measurement and errors in posting and if they were taken into account there would be more money due Claimant. Further, that a higher confidence level should have been used — 97.5%. *127He also said that the number of sites fluctuated from time to time and that this was a potential source of error in the audit. He was unable, however, to translate these potential errors into dollar figures.
On cross-examination, Dr. Benishay admitted his experience was basically in market research and econometrics and not auditing. He agreed that errors in counting might include an equal number of undercounting and overcounting.
Dr. Arken, called by Respondent on surrebuttal, testified that statistical techniques used in econometrics and marketing opinion polls are different than those used in auditing and that in auditing they do not exceed the upper confidence level of 95% and are usually between 90 to 95% upper confidence level. However, if a 97.5% upper confidence level were used, Claimant would have $8,118.00 additional monies due. As to errors in measurement, the same was not a problem in statistical sampling audits because the errors in measurement have been built into the formula and are accounted for. He said, “If the errors are random errors and do not reflect a deliberate bias, and there is no evidence of that in this case, it is obvious that the average of those errors would be equal to zero or something close to it”. Stated another way, without deliberate bias, it is just as likely that a monitor would over count as under count. Dr. Arken maintained that the criticisms of Dr. Benishay were invalid and proceeded to explain at length his reasons for this statement.
Claimant argues that the state of the art of statistical sampling is not well enough developed to be used as a claims adjustment tool. In support, Claimant cites testimony by Dr. Arken that there was a one chance in 20 that the figures developed were wrong and that this average *128is insufficient to deny Claimant’s claim. We disagree. No court requires evidence that is 100% certain of truth. To adopt a standard suggested by Claimant would deprive the courts of most evidence. Most courts would be well satisfied with evidence that does not approach a 19 out of 20 certainty of truth.
Claimant notes the disagreement of the experts called in this case and argues that such disagreements illustrate confusion in the statistical community. We disagree. Mere disagreement of experts does not invalidate a mathematical formula application. We find that the testimony of Dr. Arken was clearly more logical and credible than that of Dr. Benishay.
Claimant argues that the monitors witnessed only $5,326.00 worth of meals they considered ineligible and that statistical sampling is not accurate enough to allow the withholding of $181,039.00 based on that small sample. Again we disagree. The most credible testimony was that there was a 95% probability that the withholding of money for ineligible meals was accurate based on that sampling. Any inaccurate counting or posting was built into the mathematical formula and accounted for. Errors in judgment in the sample were resolved in favor of Claimant. The conclusions of the audit report were founded on generally accepted accounting procedures.
This Court is further guided by the case of State of Georgia v. Califano (N.D. Georgia, 1977), 446 Fed. Supp. 404. This case involved a suit by the State of Georgia for reimbursement of monies paid by the State to doctors who had provided services to Georgia Medicaid recipients. It was defended on the basis that an audit conducted by the use of random statistical samples was invalid. The court said:
*129“The court concludes that the use of statistical samples was not improper. Projection of the nature of a large population through review of a relatively small number of its components has been recognized as a valid audit technique and approved by courts in cases arising under Title IV of the Social Security Act. (Cites omitted.) Moreover, mathematical and statistical methods are well recognized as reliable and acceptable evidence in determining adjudicative facts. (Cites omitted.) However, to find that statistics may be admitted as evidence of a proposition is not to say that the statistical model will always be conclusive. The weight which must be given to such statistical evidence is not necessarily one which must be considered by the fact finder in light of the practical difficulties in obtaining a claim by claim review! In the instant case, statistical sampling was the only feasible method of audit available to HEW . . . Audit on an individual claim by claim basis of the many thousands of claims submitted each month by each state would be a practical impossibility as well as unnecessary . . .”
Although the Georgia case differed in several important respects from the instant case, the logic of the reasoning as above quoted is compelling and could be transferred to the instant case.
So also is the decision in Illinois Physicians Union v. Miller (7th Circuit, 1982), 675 Fed.2d 151, in a claim challenging the procedures of the Illinois Department of Public Aid in auditing physicians who are reimbursed for medical expenses under Medicaid. The court held that the use of a statistical sampling and extrapolation auditing procedure is not arbitrary, capricious or discriminatory where there is an opportunity to rebut the initial determination.
In the instant case, it is the opinion of the Court that the audit report, while not conclusive, is admissible and valuable probative evidence. Claimant, although able to show a few instances of mistake, was unable to demonstrate that the samples taken were substantially incorrect. The few errors shown by Claimant were adequately covered by the formula used.
This commissioner further disagrees with Claimant’s contention that Claimant was never advised that a statis*130deal sampling audit would be used as a claims adjusting tool and that had they been informed they would not have assumed the risk of joining the program. It is agreed that neither the contract nor the handbook authorized or informed the reader of such type of audit. The evidence was conflicting as to whether the matter was fully explained at the training session. Rabbi Jerry Miller, later a representative of Claimant, attended the session as did his brother, Glenn Miller, both of whom later administered the program for the Claimant. Rabbi Miller denied being told that statistical sampling would be used for the purpose of ascertaining reimbursement. He was supported by Claimant’s witness, George Hanlon, a deputy commissioner of the Department of Human Services of the City of Chicago, who also attended the training session. Hanlon could not recall that the participants in the session were told that statistical sampling would be used as a claims adjustment tool.
However, Respondent’s witness, as to this training session, was Larry Hartón, an auditor from the U.S. office of audit, who testified that he conducted a two-hour session at which, among other matters, he explained that his office would be conducting statistical sampling of the program including the amount of eligible meals served. He explained that for the first two weeks of the program the statistical sampling would not be used for the purpose of projection, but for the purpose of giving sponsors the opportunity to work out any bugs in their program with regard to the amount of meals ordered, accuracy of data, record keeping and required components of meals. He testified that Rabbi Miller voiced unconcern over the statistical sampling audit method, having had experience the previous year with this method while he administered a similar program for a different sponsor.
*131It is this Court’s opinion that Rabbi Miller must have known that neither the USDA nor the OEA had monitors on duty every day at every site. What other practical way would there have been for the USDA to protect itself from inaccurate or inflated claims? Moreover, the audit report is merely evidence, not requiring advance notice to a claimant.
Equally unavailing is Claimant’s argument that Claimant has no contractual right to pass on all of its losses to its vendor. Although the contract between Claimant and Open Kitchens, Inc., does not give Claimant the right to refuse payment to Open Kitchens, Inc., on the basis of statistical sample audits, Claimant would have the same right, in litigation between itself and Open Kitchens, Inc., to use as evidence the statistical sampling audit as does Respondent in this case.
Claimant argues that even if the audit was correct, Respondent’s only remedy pursuant to the contract was to terminate Claimant as a sponsor. We disagree, in that the contract provided as follows in Paragraph C:
“The Service Organization agrees to . . .6. Claim reimbursement only for the type or types of meals specified on the Site Information Sheet served to children during the approved meal service period at sites and account separately for any meals served to program adults.”
and further that Paragraph F of the contract relating to termination of the sponsoring organization is permissive, using the word “may” and does not imply that such is the exclusive remedy. Claims for ineligible meals constitute a breach of the contract, the remedies for which are set by general law and need not be spelled out in the contract.
Claimant also argues that all witnesses agreed that Claimant’s duty upon receiving information as to problems was to take corrective action and that where a sponsor does take corrective action it is doing all that it *132can reasonably do and therefore should not be “punished” by the use of the statistical sampling audit. While the extent of Claimant’s corrective action is disputed, it is evident by the audit that substantial corrective action was not fully taken at every site. The use of the statistical sampling audit is not punishment. It is merely one method and perhaps the only practical one of ascertaining the eligibility of meals. While Claimant did have the duty of taking corrective action, this duty was not the only one it had under the contract. It had the further duty of refraining from claiming ineligible meals.
More vexing, however, are Claimant’s contentions that the inherent difficulties in the program and certain deficiencies in the activities of the IOE as outlined in the audit report require an assumption of the risk by the Respondent.
As earlier outlined, the largest discrepancy in the figures of each party are in the area of leftover meals. This reflects the difficulty in predicting in advance the amount of meals necessary at a site while at the same time being certain that no child attending the site will be deprived of a meal. The sites were approved by IOE and the initial determination of the number of meals for each site was approved by IOE (although the responsibility for ordering correct amounts was the responsibility assumed by Claimant).
The audit report stated that:
“Some of the deficiencies in this report could have been significantly mitigated if the SA (IOE) had distributed sponsor applications earlier in the year. This would have provided the SA (IOE) immediate access to potential sponsors. This, along with effective management and staffing would have ...(c) enabled the SA (IOE) to timely monitor feeding site operations which probably would have reduced the ineligible meals . . . .”
Thus, the very audit report which constitutes the defense *133of Respondent points a finger at Respondent as sharing in the cause of the nonperformance. Therefore, we feel that Respondent, being somewhat culpable, should share somewhat monetarily in the loss occasioned by the ineligible meals. While Claimant’s theories of defense and evidence did not include a theory of sharing of risk, we feel that the audit report and the surrounding facts should be considered in a light most favorable to Claimant.
Accordingly, we think that the figures should be adjusted so that an upper confidence level of close to 100% be used. Using the figure of $8,118.00 testified to by Dr. Arken as being the additional money to be due Claimant at the confidence level of 97.5%, a doubling of that figure should be considered by the Court as approaching 100% confidence level and as being owed by Respondent to Claimant to remove all doubt as to the reliability of the result. Thus, it is hereby ordered that Claimant be, and hereby is, awarded:
(a) $38,588.00 being the amount claimed by Respondent to have been paid but for which Respondent failed to sustain its burden and,
(b) $16,236.00 being the amount necessary to remove all potential doubts and errors; for a total award of $54,824.00.
Claimant’s claim for interest is denied.