much less than it was in the estimates Barger used. Thus, the level of our pay roll series in all years, as well as the movement since I939, is independent of Census data except for a small part of the total. Social security data, the principal source of information on compensation of employees, is not used in arriving at the final product aggregates. In fact, the statistical interdependence between the annual Commerce income and product estimates, aside from certain wash items, like domestic service pay rolls which appear on both sides of the account, is largely restricted to the movement of some major portions of the two aggregates prior to I939 and to some overlapping of sources as between the estimates of personal consumption expenditures and those of income of unincorporated enterprises. This is not, of course, to deny the probability that many firms report the same figures to various collecting agencies, even though they may be erroneous. Neither does it deny that the investigator, when in doubt as to the best estimating procedure or faced with conflicting evidence, properly examines any related data referring to the opposite side of the account for supplementary guidance.