LUMBARD, Chief Judge:

■ The petitioner appeals from a Tax Court judgment which sustained the Commissioner’s determination of income tax deficiencies for 1953 and 1954. The petitioner’s decedent, Lillian Virginia Sperling, filed joint returns for those years with her late husband, Eugene Sperling, a dealer in precious metals. The petitioner contends that the Tax Court erred in disallowing credit for certain alleged purchases of metals and in computing the tax liability by the cash method, which had been used by Eugene, rather than the accrual method. We affirm.

The present controversy may be attributed in large part to the black market which existed in some precious metals during this period. Persons engaged in black market transactions are understandably reluctant to have their activities documented in detail, and many of Eugene’s customers insisted on cash dealings. Most of Eugene’s checks were made out to cash and indicate neither recipient nor purpose, and little additional information can be gleaned from his books. The Tax Court found that they “show little more than cash receipts and disbursements. In some instances the name of the vendor in a purchase transaction is indicated, but in many instances a credit to purchases is explained by only the notation ‘cash’ or ‘sweeps,’ etc.”

The first issue stems directly from the inadequacy of the business records. To determine the purchase credits which should be allowed, checks made out to cash were matched against bills for metal. Where a bill and check could be matched as to date and amount, a credit was allowed. The petitioner argues that credit should also be allowed for two additional bills which cannot be matched with any single check but which do approximate in amount certain combinations of checks. We do not think that it was clearly erroneous for the Tax Court to reject this argument.

This is not a case in which it is reasonably certain that the taxpayer had additional expenses but the amount cannot be properly documented. Compare Cohan v. Commissioner, 39 F.2d 540 (2 Cir. 1930). The proceeds of these cheeks may have been used to purchase metals or for personal expenses. The fact that in combination they approximate but do not equal bills for metal does not make out a compelling case for the former alternative.

The petitioner’s second argument is based on the records of Kastenhuber & Lehrfeld (K&L), a refiner of precious metals with which Eugene dealt during this period. From K&L’s ledger account for transactions with Eugene the following summary was prepared:

The cash debits and credits represent respectively payments by K&L to Eugene and payments by Eugene to K&L. SimL larly, the metal debits and credits represent metal transferred to and from Eugene. The processing expenses were not fully explained by the petitioner’s witnesses, but they appear to be expenses *203 incurred by K&L in processing metal for Eugene’s account.

The petitioner’s position with respect to these figures has shifted somewhat during these proceedings, 1 but it appears now to be as follows: First, the K&L cash and metals accounts show a net credit of $17,709.32 for 1953 and a net debit of $11,789.36 for 1954 and, conversely, Eugene’s net income should be decreased for 1953 and increased for 1954 by these amounts. Second, the net processing expenses — $7,552.94 for 1953 and $6,248.61 for 1954 — should be credited to Eugene’s cost of goods sold.

There are at least three flaws in this line of reasoning. First, Eugene’s own books, from which the tax returns were prepared, also refer to transactions with K&L; they show net sales to K&L of $28,503.68 for 1953 and net purchases from K&L of $33,405.27 for 1954. Thus, substitution of the K&L ledger figures would reduce Eugene’s 1953 net income by $46,213.00 but increase his 1954 net income by $45,194.63. This result is not only substantially less favorable to the petitioner than that suggested in his argument, but it casts considerable doubt on the validity of his entire line of reasoning.

Second, the petitioner’s claim for processing expenses, though offered as an additional claim, seems necessarily to be subsumed in the claim just discussed. Eugene was entitled to credit for processing expenses only so far as he. paid for them. The debit entries in K&L’s ledger certainly do not themselves reflect payments by Eugene. Presumably the expenses were passed on to Eugene, but his payments would appear in the above tabulation only as part of the general cash credit, which we considered in regard to the petitioner’s previous claim.

1. Finally, the fact that the K&L books are kept on an accrual basis while Eugene used the cash method is sufficient explanation for some difference in result for any particular period. Eugene’s net cost in dealings with K&L for 1953 and 1954 would be increased by only about $1,000 if the K&L records are substituted for Eugene’s own records. This discrepancy — a fraction of one per cent of their total dealings— provides no basis for assuming that Eugene’s records understate his costs.

The preceding point leads to the petitioner’s principal contention: that it was improper to compute the estate’s tax liability on a cash accounting basis even though Eugene had used this method in keeping his books and filing his tax returns.

The petitioner’s argument is based on § 446 of the Internal Revenue Code of 1954 and on regulations promulgated under it. 2 First, § 446(b) of the Code and Reg. § 1.446-1 (a) (2) bar the use of an accounting method which “does not clearly reflect income.” Second, where “the production, purchase, or sale of merchandise of any kind is an income-producing factor,” account must be taken of beginning and closing inventories. Reg. § 1.446-1 (a) (4) (i). Third, “in any case in which it is necessary to use an inventory the accrual method of accounting must be used with regard to purchases and sales” unless the Commissioner authorizes use of another method. Reg. § 1.446-1 (c) (2) (i).

Eugene derived income from the purchase and sale of merchandise, and the Commissioner did not authorize use of another method. (We do not regard the Commissioner’s earlier failure to object to use of the cash method as equivalent to authorization of its use.) It follows that the Commissioner could properly have used the accrual method *204 in determining the tax deficiency. And we assume for purposes of this appeal that it also follows that the petitioner could have responded to the Commissioner’s determination of tax deficiencies by demonstrating that there would be no deficiency, or a smaller deficiency, if the liability were redetermined using the accrual method.

However, the petitioner has made no attempt to show even that use of the accrual method would reduce rather than increase the deficiency. The question presented is whether the petitioner can successfully oppose the Commissioner’s determination simply by pointing out that the Commissioner has worked within the framework of the taxpayer’s own, improper accounting system. We hold that, on the facts of this case, this is not enough.

It is difficult to find any basis in reason for thus placing the burden on the Commissioner to place the taxpayer’s tax returns on a different accounting basis. As this case comes to us, Eugene’s 1953 and 1954 tax returns show two defects. The first is that certain alleged credits for the purchase of metals are not adequately documented. The second is the use of an improper accounting method. What the Commissioner has done is to determine deficiencies attributable to the first defect but to leave the second alone. We may guess that this decision was made because it was believed that the game was not worth the candle: the additional deficiencies, if any, which use of the accrual method might reveal would not warrant the considerable effort which a redetermination would require. Whatever the reason, however, we cannot see why the decision should be held improper.

The petitioner has cited no case directly supporting his contention that this burden should be placed on the Commissioner. Many of his citations merely uphold the Commissioner’s power to require use of the accrual method. The rest, so far as they are relevant, bear on two related but distinguishable situations.

First, where a taxpayer establishes that the Commissioner’s determination is erroneous, he need not also establish what is the correct determination. See, e. g., Grubb v. Commissioner, 315 F.2d 753, 759 (6 Cir. 1963). However, this rule necessarily must be limited to cases in which the error infects the deficiency asserted by the Commissioner. In this case the deficiency is based on the petitioner’s inability to document certain alleged costs, and the alleged “error” is the Commissioner’s failure to redetermine the liability using the cash method. The only apparent relation between the two is that the difference between the cash and accrual methods is one reason for regarding the K&L ledger account as unsatisfactory evidence of Eugene’s net costs. It is not the only reason, however, or even the principal reason; it is not by itself sufficient to place the burden on the Commissioner to determine what is, at best, a counter-claim by the petitioner.

A second line of cases cited by the petitioner stands for the proposition that the Commissioner as well as the taxpayer is required to use reasonable accounting methods. Contrary to the instant case, however, those cases arose from the substitution by the Commissioner of his own accounting method for that of the taxpayer, either generally, Jeff Rubin, 13 T.C.M. 1094 (1954), or as to treatment of a particular item, Keneipp v. United States, 87 U.S.App. D.C. 242, 184 F.2d 263 (D.C.Cir. 1950), or where the taxpayer has, in effect, had no system at all, Elmer D. Willis, 16 T.C.M. 1013 (1957).

Of course, it is possible that the taxpayer’s accounting system may be so inadequate that the Commissioner has no choice but to redetermine the liability on the basis of his own system; this apparently was the case in Elmer D. Willis, supra. But where the taxpayer does have a rational, though improper, system, we think that the Commissioner may choose to let it stand and to pursue only specific deficiencies within it. And so long as the deficiencies which he de *205 termines are not themselves infected by the system’s shortcomings, the taxpayer may not avoid the deficiencies simply by pointing to the inadequacy of his own accounting system.

The judgment of the Tax Court is affirmed.

Notes

. This has been in part due to the fact that originally four years — 1951 through 1954 — were at issue. Examination of the returns for the two earlier years, however, was held by the Tax Court to be barred by the statute of limitations.

. The 1953 return is governed by the 1939 Code, but the provisions of the two codes and of the regulations thereunder are substantially similar so far as relevant to this appeal.