多数の計数者による赤血球数算定誤差の実験結果について

抄録

1. Erythrocytes in a sample of diluted blood (201 times) was distributed among 9 organs of research and were examined by 28 investigators belonging to these organs, in a way that erythrocytes in one of 20 partitioned medium squares (1/5×1/5mm².) on a hemacytometer were to be counted respectively. And the results were found as follows: (1) The variance within the same investigator agreed with the theoritical value of the variance deduced from the binomial distribution, while it did not with that from the Poison distribution. (2) The variance between positions of the squares on the hemacytometer was not significant as against the random error. (3) The variance between investigators within the same organ of research was significantly greater than that within each investigator, and the variance between different organs was significantly greater than the former, i.e. that between investigators within the same organ. (4) The mean erythrocyte count was 4.512 millions in 1 cub. mm. of blood in this sample. And on this condition the coefficient of variation (c.v.) was 4.7% in the case that same investigator counted erythrocytes in the 5 medium squares repeatedly with the same hemacytomerer, it was 5.7% in the case that each of several investigators in the same organ counted erythrocytes in the 5 medium squares once with the same hemacytometer respectively, and finally it was 7.1% in the case that each one investigator from different organs counted erythrocytes in the 5 medium squares with his own approved hemacytometer respectively. (5) It happened in 42% of the total cases that, corresponding nearly to the theoretically deduced value, that the range between the maximum and minimum erythrocyte counts in each of 5 medium squares was over 20. This proved the rule that the range must be within 20 as non-sense. 2. No change was recognized in the erythrocyte counts even when the diluted blood in a test-tube was left for 53 hours.