204 PROFESSOR LETTS AND N. COLLIE ON THE 
phosphorised benzyl-cacodyl have an almost identical composition; and we 
also noticed that dibenzyl-phosphine would require similar numbers. 
Thus— 
Calculated for 
TA 
(C,H) ,P5>' (CAH,) PO... (C,H) FHP: 
Carbon, 1 .%ea 2 OG tat .' 18°9 
Hydrogen, . : 661), GG 5 | 66 
We did not think it likely that dibenzyl-phosphine had been formed in the 
reaction, as we could not account for the hydrogen atom which it requires ; 
but bearing in mind the results of our experiments on the action of baryta on 
the acid sulphate of tetrabenzyl-phosphonium, it did not appear impossible 
that oxide of tribenzyl-phosphine had been formed ; for, by the action of phos- 
phide of sodium on water, caustic soda is produced: this might react on 
chloride of tetrabenzyl-phosphonium, and give rise to oxide of tribenzyl-phos- 
phine and toluol. 
At first sight, such a supposition may not appear probable, as haloid salts 
of methyl- and ethyl-phosphonium are not changed by caustic alkalies; but we 
have shown that corresponding salts of benzyl-phosphonium possess very 
different properties from these bodies. On treating the product of the action 
of phosphide of sodium on chloride of benzyl with water, abundance of phos- 
phuretted-hydrogen was evolved, showing that a considerable quantity of 
phosphide of sodium had remained unacted on. The solution was boiled; and 
thus, if alkalies really act on chloride of tetrabenzyl-phosphonium in the manner 
we have indicated, we have the necessary conditions for the production of oxide 
of tribenzyl phosphine.* 
As a further argument for supposing that the oxide had really been 
obtained, and not the cacodyl, it will be noticed that, although the percentage 
of carbon calculated for the two bodies varies by only 0°8 per cent., the results 
of our analyses are more favourable to the supposition that the body is the 
oxide, and not the cacodyl. For we obtained 0°3 per cent. too much carbon 
for the cacodyl, and therefore 0°5 per cent. too little for the oxide; and in 
carefully conducted organic analyses the carbon is often too low, but seldom 
too high. 
We had noticed that oxide of tribenzyl-phosphine (obtained as described 
at p. 198) combines with iodide of zinc to form a compound (analogous 
to Hormann’s zinc iodide compound of triethyl-phosphine oxide) of charac- 
teristic crystalline form. If, then, the substance were the phosphine oxide, the 
* We have since proved that alkalies act very readily on chloride of tetrabenzyl-phosphonium, On 
boiling a solution of the chloride in alcohol with potash or soda for a few minutes, decomposition occurs, 
chloride of the alkaline metal is precipitated and the solution contains oxide of tribenzyl-phosphine, 
which is easily identified by its insolubility in water and other characteristic properties. 
