Ein statistisches Verfahren zur qualitativen Bestimmung der Immunität mittels Hautreaktionen

A statistical method for the detection of immunity by means of the rabbit skin test. This paper discusses the statistical basis of a technique that has already, without detailed explanation, been applied in a previous paper by Dr Winsser (1). 5 plots, arranged in 5 rows (25 plots in all) laid across the skin of a rabbits back, are injected with infectious matter in amounts inc easing in geometric progression. To four of these rows sera prepared from four different test persons are added; and the immunity is judged by the resulting decline in skin reaction as compared with the row, mostly situated at the caudal end, which serves as control. Since at the lowest concentrations of infectious matter the skin reactions were altogether too weak to show any marked decline, these were ignored. The sum, y, for each serum of the reaction values at the remaining four concentrations is compared with the corresponding sum, yk, for the control, and for the difference (yk‐y), 5 % and 1 % levels of significance are determined. To determine these limits, variations in sensibility between animals and over the area of skin of each individual had to be taken into account. This was done by analysing the skin reactions of 5 animals infected without serum; the use of more animals being considered too costly. The results show significant differences between individual means, insignificant differences between individual regressions along the back, and a significant common regression. In calculating the 5% and 1% percent limits it was assumed that the residuals with respect to individual regression lines are normally distributed, which they clearly are not; and that the individual regressions are equal, which, despite the insignificance observed, is probably not correct. To atone for these deficiencies the limits imposed were more severe than those commonly adopted. Some of the criticisms that may be directed against this procedure are recognized and discussed.