W. J. Bair June 18, 1979 Page 3 Also, the soil data in Appendix D are inadequate to allow us to double check their dose calculations. Data units are not identified, there are too many significant digits, and the error terms are not defined. 3) Lack of any error analysis on final dose estimates. Robison indicates this will] be forthcoming at a later time, but I include it here to emphasize how important it is that this type of analysis be conducted. In fact, I believe the whole approach used in the paper should be directed toward evaluating the probability that a given individual will exceed guidelines. difficulties. Admittedly, this approach is fraught with Nevertheless, I recommend that the authors begin thinking in those terms so that some future version of the paper can reflect the probability approach. 4) Tables 1, 2 and 3 (giving initial dose rates, concentration ratios and diets} are deficient in that absolutely no attempt is made to quantify or even mention the range of errors in the data or to indicate the number of samples involved. I have consistently raised this objection to past LLL dose assessments, but to no avail. - We should insist that future LLL reports follow established scientific principles and indicate whenever possible the limits of error on these types of data. 5) Others in the Advisory Group are more qualified than I to evaluate whether or not the specific models used to obtain doses are satisfactory for the Enewetak situation. However, I think the authors should indicate those instances (if there are any) where these models have in any sense been validated. For example, has Bennett's bone model been tried on data other than that for which it was initially developed? If so, how did it perform? as to methodology. estimated? Bennett's (1973) paper is quite sketchy For example, how were the parameters c, g, and 2% A more crucial question is whether LLL used Bennett's estimates of c, g and A in the Enewetak calculations. They could be severely criticized if that is the case since Bennett's estimates are based on New York and San Francisco data. The applicability of the parameter estimates used should be established in the paper.