Half‐loudness adjustments have been obtained from 18 observers at 10 intensity levels from 10 to 100 db. Adjustments were made with two methods. An analysis of variance of the attenuation scores necessary for half‐loudness showed: (1) Intensity accounted for a quarter of the variance. This variance indicates the extent to which the attenuation necessary for half‐loudness varies with different intensities. (2) Consistent differences between observers accounted for 35 percent of the variance, indicating that some observers consistently require much greater or less attenuation for half‐loudness than others. (3) One‐tenth of the variance was a result of the fact that the shape of the half‐loudness function with respect to intensity was different for different observers. (4) The error variance, caused by repeated adjustments by the same observer at the same intensity and with the same method, was 22 percent of the total. (5) Other variances were practically insignificant. The average half‐loudness function differs markedly from previously reported results. It is suggested that this discrepancy, as well as the large differences between individuals, is largely a result of the fact that observers have difficulty establishing the correct fractional value, and thus are extremely susceptible to indirect suggestion. A means of avoiding this problem of a nonvalid fraction is suggested.