84 Journal of Agricultural Research voi.xii. No. a 



cent of the extract (equivalent to about 0.02 per cent of gossypol in the 

 diet) and found that the growth of the rats was greatly retarded. Our 

 experiments have led to the conclusion that raw cottonseed kernels are 

 highly toxic to rats, but that cooked cottonseed is only slightly toxic. 

 Whether cottonseed meal made from cottonseed sufficiently cooked 

 with moist heat is toxic to rats seems to depend on the diet in which it 

 is fed. In a short feeding experiment a diet such as was used by Rich- 

 ardson and Green (ii) (45 per cent of cottonseed meal, 17 per cent of 

 whole milk powder, 10 per cent of starch, and 28 per cent of lard) has 

 shown no definite toxic effect ^ on our rats even when it contains a 

 short-cooked (28 minutes) meal. When the meal is the sole source of 

 vitamines, protein, and minerals, we have rarely had such favorable 

 growth as is reported by Richardson and Green (//) and by Osborne 

 and Mendel (/o). 



If compared with ether-extracted raw cottonseed or with soybean 

 meal, the rate of growth has been very small. The explanation of this, 

 according to Osborne and Mendel, might be that the diet was unpalatable 

 and that consequently less food was ingested. Unpublished experi- 

 ments indicate to us that there still remains something toxic in long- 

 cooked cottonseed meals which in restricted diets is objectionable to 

 rats, causing a lower food intake, but the effect of which is overcome in 

 supplemented diets. 



This seems to be the same phenomenon discussed by McCollum (7), 

 who finds that in diets very well supplemented the toxic effect of the 

 fat of wheat embryo and other slightly toxic substances is overcome. 

 The evidence of a toxic factor of moderate intensity for rabbits and 

 pigs is also furnished in our rabbit and pig experiments. 



TOXICITY OF RAW COTTONSEED KERNELS 



In a large number of experiments with rats we have found that the 

 effect of cottonseed products on rats can be predicted accurately when 

 chemical tests indicating the presence or absence of gossypol have been 

 made (Table I). In order to appreciate the significance of some of the 

 experiments we may briefly describe the properties of gossypol. A 

 yellow plant pigment having the apparent formula C30 Hjg O9 (molecu- 

 lar weight 532), not soluble in and not extracted by petroleum ether, 

 readily soluble in acetone and ether, moderately soluble in alcohol, 

 benzene, and chloroform, dissolves readily in sodium hydroxid (NaOH) 

 and sodium carbonate (NajCOg) and is slowly soluble in sodium bicar- 

 bonate (NaHCOg). It may be titrated as a dibasic acid with aqueous 

 alkalies. It crystallizes well from a mixture of ether and acetic acid as 

 a sparingly soluble substance containing 10. i per cent (i molecule) of 



'Data ia article not yet published. 



