212 PROFESSOR LETTS AND N. COLLIE ON THE 
flask connected with a reversed condenser. 100 grms. of redistilled xylol 
(boiling point 135°-137° C.) were then added together with 1°3 grm. of sodium. 
A current of hydrogen was then passed through the apparatus, and the mixture 
kept gently boiling. When most of the sodium had been acted on (which 
required some days’ digestion), the xylol was poured off and filtered. On 
cooling, it deposited an abundance of crystalline matter. This was collected 
on a cloth filter, well squeezed to free it from adhering xylol, and dissolved in 
boiling alcohol. On cooling, crystals separated having the appearance of oxide 
of tribenzyl-phosphine, and which were proved to consist of that body. The 
xylol from which this oxide had separated was distilled to dryness, and the 
residue taken up with boiling alcohol. The solution on cooling deposited beau- 
tiful silky needles, which were recrystallised twice from alcohol. In spite of 
their very different appearance from other specimens of oxide of tribenzyl- 
phosphine, a most careful examination showed that they consisted of that body 
' (see p. 205). We are completely unable to account for the difference in appear- 
ance of the two quantities of the oxide obtained in this experiment. No one 
would imagine that they were the same body. We could not obtain any other 
definite products from this experiment. 
Now the production of the oxide may be accounted for in two ways—(1) 
the chloride of tetrabenzyl phosphonium was not perfectly dry, and caustic 
soda was formed, which then acted upon it (as we have already shown), to give 
toluol, common salt, and the oxide; (2) tribenzyl-phosphine was formed, and 
absorbed oxygen from the air during the subsequent processes to which the pro- 
duct of the reaction was submitted. We have repeated the experiment several 
times, using every precaution to prevent water or oxygen from coming in con- 
tact with the mixture of sodium, xylol, and the phosphonium chloride, but 
always with the same result—viz., production of large quantities of the oxide. 
At present we do not know which of the two explanations we have given of its 
production is the correct one. 
We may mention that finely divided silver acts on the chloride of tetra- 
benzyl-phosphonium when the two are heated together ; the action, however, 
only occurs to a slight extent, and we were not successful in obtaining any 
definite product. 
The other method that occurred to us for obtaining tribenzyl-phosphine 
from the chloride of tetrabenzyl-phosphonium was to treat the latter with 
phosphide of sodium, which we hoped would react so as to give tribenzyl. 
phosphine and chloride of sodium, 
3(C,H,),PCl+ Na,P =3NaCl+4(C,H,),P. 
The following experiment was made :—3 grms. of phosphide of sodium and a 
little xylol were heated in a sealed tube, with 13 grms. of chloride of tetrabenzyl- 
