In 1917 the Chase National Bank caused Chase Securities Corporation to be organized for the purpose of dealing in securities, that being á type' of business in which the Bank itself could not legally engage. The Bank provided capital for the affiliate by declaring a special dividend in its favor in return for the issuance by the affiliate of *983all its capital stock pro rata to the stockholders of the Bank. The stockholders surrendered their certificates and received in return instruments with the certificates of stock in the Bank and the Securities Corporation printed on opposite sides.
Under the agreement contemporaneously made the stock of the Bank and that of the affiliate could not be transferred or dealt in separately, but only in units comprising one share of stock of each corporation. This agreement might be modified or terminated only by vote of a stated percentage (more than a majority) of the shares both of the Bank and of the affiliate. While the arrangement persisted, market quotations in respect of the stock were always in terms of a share of stock of the Chase National Bank. Dividends of the two corporations were paid as a unit.
In .March and April of 1930 appellant purchased .on the market 300 units of this stock at a total cost of $50,696.25. It did not then allocate, or attempt to allocate, its investment as between the two stocks.
The Federal Banking Act of 1933, 48 Stat. 162, required the separation of national banks from their securities affiliates. On June 14, 1934, in compliance with the Act, the existing arrangement between the Bank and its affiliate was terminated and the certificate of incorporation of the affiliate changed to eliminate the restriction preventing 'the separate sale or transfer of its shares. The old certificates were then surrendered and a physical separation of the shares effected. On and after June 15, 1934, the stock of the affiliate was for the first time quoted separately from that of the Bank.
On December 22, 1936, appellant sold its 300 shares of Bank stock, but retained its 300 shares in the affiliate. In its income return for that year appellant did not claim a loss on the- transaction, but subsequently (through the medium of a claim for refund on account of overpayment of its 1936 tax) it asserted a loss of $22,055.23. The amount of loss was arrived at by assigning to the Bank stock 70% and to the stock of the affiliate 30% of the original cost. The claim for refund was rejected, and on a suit to recover the alleged overpayment judgment went against appellant. The court found that the original investment made in 1930 was a single investment, and that it is neither reasonable nor practicable to allocate to each stock a portion of the amount paid for the two as a unit.
There is no statute or regulation providing specifically for an apportionment of cost under these circumstances. The Government contends that, since the investment was single, the stock of the affiliate as well as of the Bank must be disposed of before any loss may be recognized. The Second Circuit, in a comparable situation, has adopted that view. De Coppett v. Hel-vering, 108 F.2d 787.1 The Government further argues that there was no sufficient proof upon which to arrive at an allocation if allocation were otherwise permissible. Agreeing as we do with the latter proposition, we need not consider the single investment theory.
The method originally employed in arriving at the 70-30 allocation was this: The taxpayer made a comparison of the stated net worths of the Bank and of the affiliate as of December 31, 1929, and, in addition, attempted a comparison of the average earnings of the Bank and of the affiliate for the period 1924 to 1929. At the trial, however, the taxpayer relied on the testimony of an expert whom it called as a witness, rather than upon its own method of computation. This witness, differing from the taxpayer, gave it as his opinion that as of 1930 75% of the value of the composite share was attributable to the stock of the Bank and 25% to the stock of the affiliate. The witness predicated his opinion on the book values of the respective shares as of December 31, 1929, and December 31, 1930, and upon the dividend records of the Bank and the affiliate.
As to the assets of the two corporations the witness relied entirely on book values. He knew nothing of their intrinsic or market value. Information in respect of earnings, so far as obtained at all, was obtained from estimates made by financial services such as Moody’s Manual, not from statements of the Bank. The Bank itself did not then publish earnings statements. Earnings of that institution for the year 1930 were not ascertainable by the services resorted to, due to a merger which rendered any estimate impracticable. Finding that the reputed earnings of the affiliate for a period of several years prior to 1930 were less than the dividends declared, the *984witness assumed that the dividends were paid out of surplus. He did not of his own knowledge know that to he true. To put it shortly, the expert relied throughout on the sort of secondhand information and partial data ordinarly available to the man in the street.
We think an allocation for purposes of claiming a loss can not be predicated on mere hearsay. Had there been competent evidence of earnings and of net worth a different situation would be presented. As is well known, book values assigned to corporate assets are often arbitrary, and their general unreliability is everywhere recognized. Only an appraisal of actual asset values would enable the net worths of the two corporations to be computed. This is not a situation where a segregation of values must of necessity be effected on the strength of whatever data is available. The loss, if any, on the investment may be computed and claimed without guesswork when the balance of the stock is sold.
Affirmed.
The holding of the Third Circuit in Commissioner v. Hagerman, 102 F.2d 281, appears contra. For a full statement of the evidence in that case see Helvering v. Commissioner, 34 B.T.A. 1158.