throw great light upon devitrification at depths, resulting in the formation of perhaps some felsites and porphyries.

Bonney (Presidential Address, 1885, p. 66) mentions "a fair quantity of earthy dust" along with globulites and belonites in a piece of glass which had been heated for three weeks in a crucible to a bright red heat. This seems to be a parallel case with the production of amorphous sulphur described above (page 37.) Prolonged heating seems in both cases to have produced intramolecular change by bringing the atoms into new relationships with one another; whereas in an ordinary case of fusion followed by rapid cooling the changes would appear to be rather of an inter-molecular nature. Time is thus seen to be as important a factor in the thermal dynamics of rocks as in ordinary dynamical phenomena. It is the conversion of the absorbed luminous solar rays into atomic kinetic energy which probably explains the formation of amorphous sulphur on the surface of a molten mass congealing in strong sunlight. The developement of very minute acicular crystals in the sheets of window glass, described on p. 65 of Bonney's Address, was probably due to the action of traces of atmospheric moisture condensed on the surfaces of the plates of glass. This probably had something to do also with the developement of the same structural character in the case of the piece of glass described on p. 66, though it is not easy to suggest how this could have taken place without knowing all the details from the beginning to the end of the experiment. The comparison of those cases with that figured by Daubrée (Etudes Experimentales, p. 171) is interesting.

M. Daubrée says of the amorphous mass ('le residue fixé') mentioned (2) in the reference to his experimental work, that it is found on analysis to be chiefly a mixture of hydrous silicates (p. 158), somewhat porous (p. 159), and that a comparison of the analyses of the ordinary glass and the decomposed glass shows that "le verre a perdu environ moitié de la silice et un tiers d'alcali, et que le nouveau silicate a fixé de l'eau," (p. 162). He also points out the zeolitic character of the hydrated silicate. On p. 179 he makes the following significant remark:

"En voyant le quartz se séparer si facilement du verre, il est impossible de ne pas reporter sa pensée sur les veines de quartz qui sillonnent les quartzites et les phyllades, et qui sont probablement formées, comme dans l'expérience, aux dépens des roches avoisinantes."

How near he came to the recognition of the potent action of water in the 'critical state,' is seen from the following passage (pp. 169, 71):

"Dans les expériences dont il s'agit, les deux tubes n'etant pas complétement remplis d'eau, le tube de verre ne peut plonger dans le liquide que par sa partie inférieure, aussi bien á l'intérieur qu'à l'extérieur. Cependant, il est toujours attaqúe avec uniformité dans toute son étendue. Ce résultat prouve

que, dans les conditions où nous avons opéré, la vapeur d'eau, par suite de la température, et de la densité qu' elle acquiert, agit chimiquement comme l'eau liquide. On entre alors dans un état de choses où la voie humide vient presque se confondre avec la voie sèche."

Dr. Percy in the first volume of his Metallurgy* has given some very interesting information about glasses and slags, which ought to be known to every petrologist; and I have seen specimens of devitrified glasses of extraordinary interest and beauty in his splendid collection. Analyses of both the original glass and of the devitrified (crystallized) nodules are

* London, John Murray, (new edition), 1875. See also his remarks (p. 54) on the "Conditions which seem to be essential for devitrification."

quoted by him. These were made independently by Kersten and Terreil. The latter analyst, by comparison of the results of the analysis of the glass with the analysis of the materials from which it was made, found that in the process of crystallization there was no loss by volatilization. Dr. Percy points out (p. 53) that the "crystallized glass may be regarded as augite in which a portion of the magnesia is replaced by soda." A comparison of the two analyses shows that the crystallized portions are richer in lime and magnesia, and poorer in silica and alumina, than the vitreous portions. This may perhaps be accounted for by the unequal mixture of the ingredients in the molten state.

It has been suggested by some writers that glass, being a mixture of silicates, may be regarded as a solution of one silicate in another. If so this may throw some light upon the devitrification of natural glasses. But that it is not a necessary condition of devitrification generally is clearly proved by the facts reviewed in this work in connection with devitrification of some elementary bodies (sulphur, phosphorus) and of silica and arsenic. (See Appendix i.) Zirkel* refers to an ingenious device of Vogelsang for illustrating the idea just referred to. Vogelsang made separate solutions of sulphur and Canadabalsam in carbon bisulphide; he then mixed the two and allowed a drop of the complex solution to concentrate on a glass slide by evaporation of the carbon bisulphide. In this way he obtained rhombohedric crystals of sulphur in a glassy matrix. Greater interest however attaches (in the author's mind) to the cases described in this work of the crystallization of sulphur out of a matrix of the sulphur-glass itself.†

In general (as Rammelsberg in his Mineralchemie points out, p. 39) a body in the amorphous' (vitreous) state is more easily attacked by reagents than in the crystalline state. As examples, he cites garnet, vesuvian, epidote, axinite; all of which furnish on melting glasses, which are decomposed by acids with the separation out of gelatinous silica. To these may be added the remarkable fact announced by Crookes and Tidy at the Birmingham Meeting of the British Association (1886), that powdered chalcedony was found to form readily silicate of lead and so to purify water in which salts of that metal were held in solution in minute quantities, while crystalline quartz sand had no such effect.

Again, in such a mixture of crystalline and vitreous ('colloid') silica as is presented to us in common flint (as it occurs in the chalk), something of the same sort of thing may be observed. Long observation of the degraded flints of the

*Die Micros. Besch. der Mineralien und Gesteine, p. 95.

+ See Appendix i.

gravels of the Bagshot country has made one familiar with a great variety of appearances and mineral characters which flints are capable of assuming, in what is usually called 'weathering." This includes the long-continued action of peaty waters or of humus-acids contained in the soil, which with ammonia form soluble azo-silico-compounds. In this way the successive layers of which the flint was built up very often around a central mass-a sponge it may be or an echinoderm-are brought out in a very marked manner by the corrosive action of solvents.* In other cases the flints appear to be so completely deprived of their vitreous constituents, that they may fairly be spoken of as quartzite. It is not impossible that some erroneous inferences have been drawn as to the origin of some of the gravels by observers not sufficiently familiar with these peculiarities. All degrees of degradation of flint may be seen in them down to masses which are so devitrified that they may even be mistaken for Sarsen-stones of the more compact variety. On the other hand it would be difficult to deny that the devitrified parts of flint may owe their mineral character in part to direct metatropic change, by the developement of a crystalline structure in the vitreous constituents of the original flint. I have specimens of flint so devitrified on both sides of cracks as to have acquired a stony character through zones several millimetres thick. (See further App. i., e.)

Turning again to the facts mentioned in this section of the present work as to the relative densities of the different allotropic forms of the same mineral, we are able to draw from the fact that the maximum density and stability of molecular structure is identified with the crystalline form the deduction that pressure is favourable to crystallization. This principle has received experimental verification in the cases of quartz (Daubrée), and perhaps of carbon in Mr. Hannay's experiments in the year 1880 (Note L). And in general it may be said that pressure tends in all cases to make any body pass from a less to a more dense condition, as is well exemplified (cf. also laboratory-work recorded in App. i.) in the case of the liquefaction of ice by pressure-a point which I have more fully discussed in its nature and consequences in my paper on the Mechanics of Glaciers. If however we consider for a moment the kind of pressure needed it may save us from some false inferences. The pressure which liquefies ice is hydrostatic pressure: that is to say, it must act upon the mass

*A fine example of this structure is to be seen in the Museum of Geology, in Jermyn Street; and corroded fragments of such 'banded flint' are by no means uncommon in the gravels of Berks and Surrey. See Roth (Allgm. u Ch. Geol. Vol. I. pp. 94-97) on Weathering of Quartz and Silica.' + Quarterly Journal of the Geological Society, February, 1883.

equally in all directions. It must in fact act by way of compression, as in the experiments of Sir W. Thomson, Helmholz, Tyndall, and others. The ice must be confined in the cylinder of the hydraulic press, or in some other way. Of course in the early stages of the process we have fracture of the mass, liquefaction at points of contact where the pressure is exerted, and regelation of the liquefied portion as it escapes into the free interstices (as in the well-known copper-wire experiment); and in this way some loss of bulk is experienced. But when the possible limits of this are reached in a closed space, a further exertion of pressure can be made to liquefy the whole mass, provided there is no escape for the liquefied parts.

Just so, in the case of the mineral constituents of a rock composed wholly or in part of vitreous or amorphous material. We have no right to reason from the facts which furnish our data to the inference that mere pressure can promote crystallization, unless that pressure be exerted upon the mass in all directions; in fact, by way of compression. We have no warrant in assuming that a pressure which crushes a rock will induce crystallization. It may act as an important antecedent condition by allowing freer access and circulation of water holding mineral salts in solution, and so prepare the way for paramorphic changes; but a pressure so exerted cannot be held to induce metatropic change. When these things are considered, it will be seen that mere deformation of rocks by pressure may have too much attributed to it, and probably has recently had too much attributed to it as a factor in metamorphism. So far as the heat developed by pressure is concerned, this is clearly adverse to crystallization; its direct effect tends in the opposite direction, rather towards fusion than crystallization. On purely physical grounds therefore we may demur to the notion that crystallization is caused directly by deformation of rocks by pressure; although it would be rash to dogmatize on this question until the advance of experimental physics has taught us something definite as to the critical state' of the passage of a body from a solid to a liquid, and the reverse. Acting in the dry way it would give us a dust, mere mineral matter in a state of molar division: acting in the presence of water, it would certainly help to prepare conditions favourable to paramorphic changes, because it would give us locally conditions (heat, water, pressure) approximating more or less to those which must have prevailed universally at the surface of the globe when the earliest crystalline rocks were formed.

That a mechanically-stable body like glass may not be at the same time perfectly rigid is shown by the well-known fact that mercurial thermometers,

if graduated soon after the tubes are filled with mercury, are liable to give after a time too high a reading (to the extent in some instances of from 1° to 2°), from the compression of the bulb by atmospheric pressure. This brings out very well the importance of the factor of time in vitrification, since it shows that even after a vitreous body has become so mechanically stable as to be highly fragile, a certain change (within very small limits) is still possible, under prolonged strain, for the relative positions of its molecules. Is it not possible that within such limits such forms as belonites may be developed? [A series of observations on thermometers which, having been graduated at once after being filled, some of which were exposed subsequently (others not) to prolonged temperatures at O°C or lower, might give some interesting results as bearing upon the theory of devitrification.]

As regards the necessity insisted upon in this work for hydrostatic compression as distinct from mere pressure, in promoting metatropic change quâ crystallization, the idea seems to have been even more strongly put by Prof. Heim of Zürich. From a recent perusal of his great work 'Untersuchungen über den Mechanismus der Gebirgsbildung' I find that that distinguished geologist emphasises, even more strongly than I have done, the necessity for hydrostatic pressure. He shows how by this means deeply-seated rocks may have been subjected for lengthened periods to a pressure far beyond the limits of their rigidity (überlastet); and that this has given them a latent plasticity which made mechanical (metataxic), and even molecular (metatropic), changes easy in the subsequent massive movements concerned in mountain-building. The possible 'critical state.' *It is not at all unlikely that certain minerals contained in composite crystalline rocks may in some cases have reached this state by the combined action of heat (tending to expand them) with great (hydrostatic) pressure preventing free expansion. This is strongly suggested (e.g.) in such flattening-out of masses of quartz as is shown in the Atlas (Tafel vii, figs 5, 6) which accompanies Lehmann's magnificent work Altkryst. Schiefergest, as well as by the intrusion by pressure of adjacent mineral particles into them. [The quartz of the larger pebble in fig. 5 shows no such metataxic change, but only signs of fracture under pressure.] It seems almost certain therefore that the flattened pieces of quartz have undergone the particular kind of deformation which they exhibit from their having passed through the 'critical state' (i.e. the state in which they were neither solid nor liquid) owing to their having previously existed in a different allotropic condition (hyaline or otherwise), which rendered the critical state possible for them, while the temperature was not high enough for the same state to be induced in the neighbouring quartz-masses, in which only crushing effects are seen. A similar explanation may possibly apply to the deformation by stretching represented in Tafel viii of the same work. It is not paramorphic but metataxic change which is there observed, for the minerals (e.g. the garnets) were evidently there previously completely individualized. The stretching-out in continuous bands of some of the minerals seems to show metataxis without disruption of molecular continuity, according to the slight variations of the mineral-composition of the bands; and this too may possibly be explained by those in which (under metataxis) the continuity of structure has been most completely preserved, having passed through the critical state. All traces of previous allotropic variation in the quartz-fragments (Tafel vii) may have been disguised by subsequent metatropy; but it is possible to understand how in some *I had no idea, when this was written, that the subject had been even touched by any experimental investigator. It is therefore very satisfactory to learn that Amagat has shown that carbon dichloride is solidified at a pressure of 900 atmospheres at 10°C, and benzene at 22° C. under 700 atmospheres. He points out that there is "a temperature above which solidification cannot be effected by any pressure; that is to say, a critical point of solidification." Comptes Rendus, July, 1887, quoted by Sterry Hunt, Chemical News, October 19th, 1888.

I draw particular attention to the crushing, and even pulverization, of the garnets in the non-quartzose layers, and to the preservation of their forms intact in the quartzose layers, as seen (e.g.) in the specimen of normal Granulit' in the Lehmann Collection in Jermyn Street Museum,

In a note on these, Lehmann says they have 'eine felsitische Beschaffenheit.