The Error of Determination of Toxicity.

The determination of toxicity is usually given quantitative expression by the statement of a minimal lethal dose. The common use of this expression in the literature of the subject would logically involve the assumptions that there is a dose, for any given poison, which is only just sufficient to kill all or most of the animals of a given species, and that doses very little smaller would not kill any animals of that species. Any worker, however, accustomed to estimations of toxicity, knows that these assumptions do not represent the truth. How widely different is the real state of affairs, however, is not, I think, sufficiently recognised. The fact that the “ minimal lethal dose,” whether calculated for unit weight, or for surface area, or on any other basis, varies widely for different species has, perhaps, led to the looseness of its definition for any one species. For the accurate standardisation, by biological methods, of drugs which are not available in chemically pure form, it is essential to establish a more accurate definition of such terms as “minimal lethal dose,” “minimal effective dose,” etc. Fig. 1 gives the results of the injection of four poisons into animals. The abscissae are proportional to the doses injected, the scale obviously differing for the different drugs, and the ordinates give the percentage mortality for each dose injected. The number attached to each observed point represents the number of animals injected for its determination. The curves represent percentage mortalities produced by the subcutaneous injection of tincture of digitalis into frogs, by the intravenous injection of cocaine hydrochloride into mice (see also fig. 2 and Table I), by the intravenous injection of echitamine into mice, and by the injection of dysentery toxin into mice, the data for the last being taken from O’Brien, Sudmersen and Runge (1924). A similar curve is given later (fig. 7) for the percentage of convulsions produced in mice by increasing doses of insulin, the data being obtained by the use of large numbers of animals. Shackell (1925) has published a number of similar curves, relating percentage mortalities to varying doses of different poisons, in a wide range of species. It is suggested that the curve expressing the percentage of mortality, or of some other limiting biological effect, produced by varying doses of a drug on animals of a certain species, shall be called the “ characteristic” for that particular drug, effect and species. Thus, the curve relating the percentage of convulsions produced in mice to varying doses of insulin, would be termed the characteristic for the production of convulsions in mice by insulin.