OP ARTS AND SCIENCES. 439 



small apparent losses to be observed in some of the determinations 

 in which a large excess of acetic or formic acid was employed, are 

 probably explicable by the tendency of the precipitate to change its 

 consistency as the amount of free acid increases, and, in very acid 

 solutions, to show an inclination to adhere in small amounts, but quite 

 persistently, to the vessel in which precipitation takes place. The 

 adherent precipitate may be dislodged with ease by putting a little 

 hydrochloric acid into the beaker to which the precipitate adheres, 

 covering and heating gently so that the acid volatilized may condense 

 upon the walls of the beaker and again run down. By then rubbing 

 the walls of the beaker a little and adding ammonia in excess, the 

 trace of residual titanium is completely recovered. This method of 

 recovery was applied in (2G) and (27). 



In preliminary experiments upon the first of the methods which 

 seemed to promise a separation of titanium and aluminum, it was 

 found that the precipitation of titanium by an alkaline orthophosphate 

 is complete in the presence of a large amount of acetic or formic acid, 

 and that, in the cold, and under conditions otherwise properly con- 

 trolled, aluminic phosphate fails to appear. To hold up the alumina 

 by means of acetic acid requires some care in the adjustment of the 

 acid and phosphate, and to redissolve the precipitate once formed is 

 a matter of considerable difficulty. Thus, to clear the solution of a 

 precipitate produced by 1.5 grm. of the phosphate of soda and 

 ammonia upon 0.1 grm. of alumina, it was necessary to dilute the 

 liquid to a volume of 250 cm.^ and add acetic acid until there was in 

 the solution at least five per cent of absolute acid ; and the addition 

 of 1.5 grm. more of the precipitant again precipitated aluminic phos- 

 phate. Formic acid, however, is more active, and the balance between 

 it and the precipitant not so delicate. When the proportion of three 

 parts by weight of absolute formic acid to two parts of microcosmic 

 salt is kejit, the aluminic phosphate does not fall ; and if, because of 

 a deficiency of acid, precipitation does take place, the liquid immedi- 

 ately clears so soon as the proportion of acid and precipitant is 

 restored. Formic acid, therefore, was used in the experiments about 

 to be described. The test solution of titanium was that employed in 

 the previous experiments (1) to (20), but, inasmuch as the error 

 of the process had not yet been remarked, the standards were deter- 

 mined by precipitating by ammonia, and arc therefore to be cor- 

 rected. The application of a correction of two and a half per cent 

 to the apparent weights of titanic oxide found by the ammonia process 

 — the mean correction for (1), (2), (5), (G), (9), (10), (13), (14) — 



