| 
176 PROF. W. J. SOLLAS ON A PROCESS FOR [May 1902, 
$10, | TiO, | A1,0, Fe,0,| Fe? | | 
| U5 1U5 2 3| Be,O, MnO CaO MgO KO NasO H:0 
Per cent : ies 
Sphewereceee 0°464 ...!0°142 |0°184 |0°008 | ...... O01 | O-11 | 0°004 | 
Magnetite hae : ? 
perpen, Teche) P| deaase) I see |Ieaccoc3 1349 | 0°607 | 
Oligoclase...... ILE) call GBS) |b scasae VA aisseaetd| Reese 157 | weseees 4) eaeke ee 5°23 
Ojan oor) pastel) con) Hema) | Gousoe | OFZ |! cocces ‘Il, donooe. || aoneda: |} _qooace 4:86 
Hornblende,..16°87 ...| 7°38 | ...... I OSLO ES gener 5°04 | 2°75 | 2°61 | (by |differe|nce) 
(Total; 100°79.) |59°832) 0:184/16°85 | 1°35 | 565 | 4°48 | 2°61 | 4°86 | 5:23 
[Analysis by Prof. Aiekel59:83) |e ecee | LOS Del ee ce: TOL | 4°43 | 2°61 | 6:57 | 244 | 1:29 
fants 
(Total: 101-03.) | 
The specific gravity, as calculated from the composition, is 2°73. Prof. Zirkel 
gives it as 2°865. The specific gravity of the specimen that I examined was 
2°724. 
Not only do my results yield a larger total amount of the alkalies 
than is obtained on analysis (10°09 per cent. instead of 8:99 per 
cent.), but the relative amount of potash and soda is reversed, the 
soda being in excess of the potash, instead of falling considerably 
below it. Quite possibly the amount of orthoclase in the rock has 
been under-estimated, but scarcely, we may fairly conclude, by more 
than 2 or 3 per cent., so that the excess of potash must be looked for 
in the oligoclase and hornblende. As a matter of fact, both these 
constituents give an obvious potassium-reaction to the flame-test. 
The next difficulty arises from the occasional association of quartz 
with a felspar of the same specific gravity as itself. In the event 
of this unfavourable conjunction, a separation of the quartz and 
felspar from the other constituents must be made within narrow 
limits of specific gravity : a silica-determination will then give data 
for calculating the relative amounts in which the associated minerals 
are present. 
The chance of substitution of ferrous iron for calcium and mag- 
nesium, and of ferric iron for alumina, may sometimes lead to error, 
and in the case of rocks from localities in which the prevalent 
minerals have not been analysed, this must be met by chemically 
determining the amount of iron in the rock, a not very laborious 
operation by volumetric methods. 
Discussion. 
The Prusipenr remarked that this was not the first time that the 
Author had made an important contribution to the improvement of 
petrographical methods. Most petrographers in this country were 
extremely indebted to him for the invention of the method of the 
diffusion-column. The method now put before the Society had 
yielded in the Author’s hands more accurate results than would 
have been expected. It was true that he had not yet brought the 
process to its full perfection, but it seemed highly probable that 
the discrepancies between the observed and the calculated results 
