altogether unprofitably so, since by the examination of a larger number of samples I hope I have been enabled to obtain some insight into the constitution of this curious silicate.

The specimens which I have examined are four in number: the two first taken in the summer of 1856 at the opposite sides of the pit, the two latter taken about six months later from one of the sides, in the immediate neighbourhood of each other. Their physical characters were somewhat different: No. I. was an amber-coloured transparent substance, bearing a strong resemblance to the gum-resin copal; No. II. was semi-opake, and more friable than the former; it had also a fainter tinge of yellow, and occurred in larger masses; No. III. bore a considerable likeness to No. II.; and No. IV. was of a ruby-red colour, and transparent, and appeared to be generally disposed in the form of layers, alternating with the ordinary allophane. The specimens were all found to contain, besides alumina, silica and water, lime, carbonic acid, and more or less iron; and these constituents were estimated according to the usual methods. Taking into consideration, however, the fact of the large and very variable quantity of water present in all the varieties of which analyses have been published, and by means of which such difficulties have been thrown in the way of arriving at any formula for the mineral, I considered it desirable accurately to ascertain the amount of loss which each specimen experienced by exposure to a heat of 100° C., in order to see whether any more definite view of their constitution might be obtained by knowing the composition of the dried substances. The loss thus undergone is not nevertheless wholly due to the escape of adherent water, for in each instance I have found more carbonic acid in the specimen than was sufficient to combine with the lime to form neutral carbonate; this I have supposed to exist as bicarbonate; but inasmuch as the carbonic acid in excess would be evolved together with the loosely-combined water at the temperature of 100° C., I have in every case deducted its amount from the number representing the total loss under those circumstances. In a similar manner the entire amount of carbonic acid was deducted from the total diminution of weight after ignition, in order to arrive at the whole quantity of water present. The small quantity of iron occurring in the three first specimens I have assumed to exist as protocarbonate, although I have not determined whether such is its real condition, since its amount appears to be far too minute to interfere with the formula of the mineral; but in No. IV., in which it exists in much larger quantity, I have ascertained that it is present entirely as sesquioxide.

The results arrived at in the analysis of the four specimens alluded to will now be given.

Loss percent. at 100°C. Carbonic acid exist

}

ing as bicarbonate 1.04

These constituents may be approximately arranged as follows:

If now we remove the carbonates of lime and iron, the looselycombined water and carbonic acid, which may, I presume, be safely regarded as extraneous to the mineral, we obtain the following numbers expressing its composition :

The actual formula deducible from these numbers is

A12 08. 8SiO3, 3(A12 03. 3HO) +21HO;

but I need scarcely say that it is my belief that the real formula of this specimen is

A12 03. 3SiO3, 3(Al2 03. 3HO),

the additional water having very probably crept in as an erroneous result. This might happen in many ways: the protocarbonate of iron, if really so existing in the mineral, would certainly by drying at 100° C. become converted into hydrated sesquioxide, by which means more water would be retained than the alumina alone required; or a slight variation in the relative

proportion of the three constituents would remove the anomaly; or the very hygroscopic nature of the dried substance would of itself be quite sufficient to account for such an excess of water; for I found that when, after drying, it was left for a day or two, even in clamped watch-glasses in a balance-case, it had regained half the weight which it had lost by drying. I may also state here, that, on the authority of Brande, the neutral hydrate of alumina sustains a temperature of 100° C. without loss of

The accidental constituents being, as in the former case, withdrawn, we obtain as the composition of this specimen,

And the formula arrived at from these numbers is

A12 03. 8SiO3, 4(Al2O3.8HO) +2HO.

If the 2HO are considered here, as in the former case, to be due to erroneous determination, arising from one or more of the

disturbing causes which I have before mentioned, we obtain the formula

Al2 03. 3SiO3, 4(Al2 O3. 3HO)

as the simplest expression in whole numbers of the constitution of this specimen.

Specimen III.

20.10

31

Water

Loss percent. at 100°C. Carbonic acid exist

ing as bicarbonate

If we again assume the water expelled at 100° C., the carbonate of lime, and the excess of carbonic acid, to be non-essential to the constitution of the mineral, the following numbers will represent its real composition:

The precise formula given by these numbers is

Al2 03. 3SiO3, 5(A1o 03. 3HO)+3HO;

or if we exclude, as before, the surplus water,

A12 O3. 3SiO3, 5(Al2 03. 3HO).

The fourth specimen, which from its appearance evidently contained some quantity of sesquioxide of iron, I examined in the

hope of finding that it was an iron substitution compound: it was selected with considerable care, and freed as perfectly as possible from the ordinary allophane with which it was interstratified. Its analysis yielded the following results:

Loss per cent. at 100°C. Carbonic acid exist

Treating these constituents as in the former cases, we obtain the following approximate arrangement :

ing as bicarbonate

•75

These results, if the former method of exclusion is again adopted, give the following as the true constituents of the mineral:

The formula given by these numbers places this specimen in close relationship with the last mentioned, the sesquioxide of iron replacing a small portion of the alumina, as will be seen by the following expression, which is that which the above composition yields:

A12 0. 3SiO3,

4(A12 03. 3HO)

A12 087

зно

+HO