amounted to 50 to 60 per cent. of the whole dose of crushed seed. In order to test whether such residues contained any appreciable amount of tutin, a fairly large amount was dried, extracted with ether, and the ether-soluble material administered to a rabbit in watery suspension by stomach-tube. It produced no symptoms whatever in a dose equal to 7.5 grm. of “husk” per kilogram body weight, so that the bulk of the tutin may be supposed to have been present in the watery suspension. When the seed was administered in this way the results shown in the table were obtained. Toxicity of Seed No. Material used. Dose per Kilogram, in Grammes. Result. Estimated Percentage of Tutin in Seed. 256c Sample I 3.7 Death in three hours 0.16 267 Sample II 3.7 Slight, if any, symptoms 270 " 4.0 Distinct minor symptoms 0.14 * This animal had been used for exp. 274, and had then had its thyroid gland removed, about four weeks before being used for exp. 295295 " 4.3 Death in about three hours 0.16 272 " 4.5 Symptoms more marked than in exp. 270 0.12 274 " 5.0 Severe symptoms, but recovered 0 11 282 " 6.5 Death in three hours 0.10 396 Green seed 6.0 Death in one hour and a half 0.125 The percentages of tutin in these experiments have been calculated from the results of previous work by Fitchett and other experiments by the writer. The standards adopted for rabbits were that a dose of 5 milligrams per kilogram produces no marked symptoms; 6 milligrams produces symptoms in one hour and a half, and is ultimately fatal; 7.5 milligrams produces symptoms in about half an hour, and death between one and two hours. Sample I of the seed was used in only one experiment, and in that case was more toxic than sample II. It occurred to me that this might be due to a difference in the toxicity of green seed as compared to ripe, for, owing to the method of collecting it, sample II contained more ripe seed than sample I. Fortunately, although it was late in the year (May), I was able to procure locally sufficient green berries to put the matter to a test. At first it seemed impossible without serious loss of time to mechanically separate the small unripe seeds from the green fleshy petals, and after separating enough to find the percentage of seed present (18.2) the whole unripe fruit was administered in the form of a watery suspension. This gave the following results: 27 grm. fruit per kilogram caused death in forty-five minutes; 14.4 grm. in ninety-five minutes; 10 grm. in two hours and a half. As 10 grm. fruit contained only 1.8 grm. seed, it followed that either the unripe seed contained a very large amount of tutin, or that, at this stage, it was also present in the fleshy petals. To decide this point it was necessary to obtain a clean sample of unripe seed, and after several attempts the following method was found to be successful: Berries were dried in air at a moderate temperature and rubbed between the fingers; by this means the seeds were isolated from the remainder. The material was then put,