787 
cible in all proportions with alcohol, and very sparingly soluble in 
petroleum ether. The solubility of all three acids in this oil is consi- 
derably greater than that in any other of the examined oils (see 
Table I); it is particularly striking that the solubility of salicylic acid 
in this oul still slightly exceeds that of benzoic acid *). 
Let us now determine the coefficients of distribution of the three 
acids between olive oil, resp. ricinus oil and water by the aid of 
the following solubilities of the acids in water: 
cinnamic acid 0.0546 gr. per 100 gr. water °) 
salicylic acid 0.223 „ OW De aR es 
Beneden Me eins, Miller pele cn LOO raad) 
Then we find: 
TABLE IV. 
| D.C» = Er substance in 100 gr. of oil _ 
“~~ gr. substance in 100 gr. of water 
Olive oil. Ricinus oil. 
cinnamic acid 23.6 138 
salicylic acid 10.9 66.4 
benzoic acid 11.6 43.2 
According to Overton benzoic acid would, therefore penetrate 
somewhat more easily into the living cell than salicylic acid, and 
will therefore also act somewhat more strongly narcotically. If on 
the other hand we had assumed the solubility of the acids in ricinus 
oil as basis of our considerations, we should have arrived at the 
opposite conclusion that the plasma wall is considerably more perme- 
able to salicylic acid than to benzoic acid, and that therefore the 
former acid would be the strongest narcotic, resp. disinfectant. 
On comparison of the coefficients of distribution of the three acids 
between the other oils on one side and water on the other side, we 
come to analogous contradictions. We shall not enter into a discus- 
sion of these data here, as they do not open new points of view. 
3. From this numerical material the following important conclu-_ 
sion may be drawn: 
1) This is the more remarkable as ricinus oil consists of glycerides of oxy-acids, 
and salicylic acid is an oxybenzene carbonic acid. The well-known rule of solubility 
holds, therefore, here again. 
2) Jur. Meyer: Z. f. Elektrochemie 17, 978 (1911). 
3) This value is a mean of the most probable data, recorded in LANDOLT— 
BORNSTEIN—RortH tables. 
51 
Proceedings Royal Acad. Amsterdam. Vol XXIII. 
