OPINION OF THE COURT
Petitioner Meyer, individually, in partnership or through incorporation, is the owner, in whole or in part, of four retail *225drug stores, a liquor store and a gift shop, all located in the Adirondack region. The gift shop ceased operations in December, 1968. Audits of these businesses by the respondent for the periods August 1, 1965 through February 28, 1969 and June 1, 1969 through March 31, 1972 resulted in additional assessments of sales and use tax of nearly $50,000, not including penalties and interest which are still accumulating. Petitioners no longer question the assessments on the liquor store.
Petitioners’ argument in this proceeding may be broken down into a contention that an audit of purchases was not necessary and therefore void under section 1138 of the Tax Law, and a further contention that the audits were permeated with errors.
Both contentions turn, to a certain extent, on the validity of petitioners’ own accounting and auditing procedures. Each drug store was divided into sections and each section had its own cash register. The Plattsburgh and Saranac Lake stores had four sections: drugs, cosmetics and jewelry, cigarettes and tobacco products and soda fountain. Ticonderoga and Lake Placid had only drug and tobacco sections. The Lake Placid store sold no prescription drugs. The cash registers were of varying capabilities. Certain registers, usually in the drug sections, had a tax key allowing for calculation of a subtotal for taxable items and showing tax due. Tax collected could be separately totaled. Other registers had subtotal keys but no tax keys so that taxable items could be segregated. The machines could not separately total taxable items at the end of the day. Still other registers were incapable of any segregation.
In view of these inadequacies, petitioners used various procedures to monitor and determine taxes collected. At some registers, sales clerks recorded sales tax collected on a memo pad. The pads were not always identified by date or cash register and in all cases the pads recorded only tax collected without recording taxable sales. At other registers no record of taxable sales or tax collected was kept and petitioners calculated the tax by applying a flat percentage, a figure less than the relevant sales tax rate, to gross sales shown on the register tapes. It is obvious that these methods were inadequately designed to determine tax collected, and worse, could not be monitored for accuracy without accepting petitioners’ word at key points, an outright violation of accepted accounting practice.
*226Subdivision (a) of section 1138 of the Tax Law allows the Tax Commission to determine tax due using "such information as may be available” if the amount of tax actually paid is incorrect. It then provides that "[i]f necessary, the tax may be estimated on the basis of external indices, such as * * * purchases”. Petitioners argue that available information was adequate for purposes of determining tax due. Thus, it was not "necessary” to examine external indices and the decision to do so was error. We need not decide whether the Tax Commission must demonstrate the inadequacy of available information before it may seek further information since we have determined that the available information in this case was inadequate and it was necessary to look to other sources.
Petitioners’ other claim, that numerous errors were committed during preparation of the two audits, is deserving of serious attention. Most difficult is the contention that the retailer’s markup on nontaxable items, particularly prescription drugs, is far greater than the markup on taxable items. Since the audits were based on analysis of purchases rather than sales, they failed to account for this variation and, thus, overstated the actual percentage of income attributable to taxable sales.
Petitioners are correct when they state that the auditors relied on an analysis of purchases to determine tax owing. An examination of the first audit (Aug. 1, 1965 through Feb. 28, 1969) reveals, however, that the figures arrived at through analysis of purchases were cross-checked with an analysis of sales records. Using cash register receipts the audit revealed an underpayment in taxes of 75.46% which yielded the conclusion that 61% of gross sales were of taxable items. The auditor then turned to an analysis of purchases which revealed taxable sales in excess of 71% of gross sales. He used the analysis of receipts to calculate tax liability so that any error in the second analysis due to variations in markup could not affect the tax liability calculation. Petitioners’ argument does not stand up after analysis of the first audit.
The second audit (June 1, 1969 through March 31, 1972) does not fare so well. The auditor did rely solely on an analysis of purchases, so that severe deviations from actual tax charges may have resulted. That does not end our inquiry, though. We still must determine whether the audit procedures were reasonable under the circumstances and whether petitioners’ proof of substantial error was sufficient.
*227We have already found that it was petitioners’ dereliction in keeping proper records which necessitated respondent’s further inquiry. We should add that the record supports the conclusion that matters were even worse during the second audit. Apparently, it was during this latter audit period that petitioners used the method of taking a flat percentage of total sales to calculate tax collected. This is, of course, imprecise, and leaves the auditor no hope of ever determining the percentage of taxable sales using sales receipts. Further, during the period covered by the second audit, petitioners "absorbed” the sales tax on cigarettes in violation of respondent’s regulations (20 NYCRR 525.6; see Tax Law, § 1133). Since no sales tax was recorded, it became all but impossible to determine taxes owing.
In the face of this imponderable mess, respondent sought some reasonable method of assessing taxes. Its duty was to choose a method "reasonably calculated to reflect the taxes due” (Matter of Grant Co. v Joseph, 2 NY2d 196, 206). This duty required rejection of petitioners’ sales records since they no longer bore any relationship to proper assessments. Respondent had no choice but to examine purchases to determine what items should have been taxed when resold. This was entirely reasonable, authorized, by statute (Tax Law, § 1138, subd [a]), and reasonably calculated to reflect taxes due. It then became the duty of petitioners to demonstrate error.
It is impossible on this record to determine the effect of variations in markup on the accuracy of the audit. The question is whether the evidence of the existence of variations in markup is sufficient to overcome the apparent validity of the audit and require further proceedings. For the petitioners’ position, there is testimony that petitioners used "loss leaders”—items sold slightly above or even below cost—to attract business and compete with area discount chain stores. Most loss leaders were subject to sales tax, but some were not. Petitioner Meyer testified that his main source of profit was prescription drugs. These, he stated, have a relatively high markup. On the other hand, some taxable items were sold at markups up to 75% as high as markups on drugs. Also, some loss leaders were nontaxable items. Further, it would be futile to try to determine the markup on cigarettes since petitioners unlawfully absorbed sales tax on this item. Markup could be based on either net profit or on net profit plus the taxes that *228should have been charged. However, since cash registers were not exclusively used for any particular items, it would be impossible to determine which sales were for cigarettes. Since cigarettes constituted a large portion of petitioners’ business, the problem is all that much greater. Finally, we note that petitioners were aware that respondent was using their purchases for the audit. They should have been prepared to carrry their burden of showing the audit’s invalidity at the hearing. Petitioners failed even to directly address the issue, much less provide evidence to establish appropriate adjustments.
On balance, we believe the second audit should stand. Petitioners failed to meet their burden of demonstrating error. The audit was facially sound and petitioners countered only with vague assertions, and did not even directly address the issue. More important, the record shows that the condition of petitioners’ records and the nature of their business practices preclude any intelligent application of markup differentials, even if such differentials were properly established.
We find no merit in petitioners’ other assignments of error. Petitioners claim they dealt with businesses whose purchases are exempt from sales tax, but they offered no direct evidence of any specific items that the auditors included as taxable when, in fact, they were sold to exempt organizations. Thus, petitioners did not sustain their burden of proving that the sale of the audited items was not taxable (see Tax Law, § 1132, subd [c]).
With regard to fountain sales at the Plattsburgh store during the first audit period, the auditor concluded that petitioners had failed to collect sales tax on soda sales and on sales over $1 (food sales under $1 were not taxable at that time). We recently held that where the taxpayer’s own failure to maintain proper records prevents exactness in determination of sales tax liability, exactness is not required (Matter of Markowitz v State Tax Comm., 54 AD2d 1023). Such is the case here and petitioners’ allegations to the contrary are not probative.
Petitioners also challenge the first auditor’s use of figures calculated by analysis of tapes and reports for the Plattsburgh store only. The record reveals that whatever were the deficiencies of the register tapes at that store, the other stores’ tapes were worse. No breakdown was possible. Thus, petitioners’ own bookkeeping procedures prevented the auditor from' *229analyzing cash register tapes in the other stores to determine the actual margin of error. Since bookkeeping procedures were essentially the same in all four stores and since all the books were kept at the Plattsburgh store, it is reasonable to assume that the margin of error was uniform. Petitioners offered no proof to the contrary.
The determination should be confirmed, and the petition dismissed, without costs.
Kane, Larkin, Mikoll and Herlihy, JJ., concur.
Determination confirmed, and petition dismissed, without costs.