power which the microscope had put into the hands of students, there was a danger of rocks being studied apart from their natural surroundings, and considered too exclusively and artificially as mere mineralogical and cabinet specimens, instead of integral parts of the earth's crust.

Unfortunately, petrologists are at variance in their principles and systems of nomenclature and classification. I am very sure that much of this variance is due to the differing histories of the science in different countries. English, French, and German observers have approached their subject from different standpoints, and international agreement has consequently been made more difficult. A further source of difficulty is the intrinsic nature of the subject. The eruptive rocks are aggregates of minerals which have separated out from eruptive magmas under varying physical conditions, and they share with organic groups or species the disadvantage (from a classificatory point of view) of passing the one into the other through connecting varieties. New types are turning up with startling frequency, and the old simple nomenclature no longer covers the ground. These discoveries sometimes prove disconcerting, for they impose upon us the duty of re-considering and often of re-adjusting our views.

In advocating a few years ago, with Mr. Petterd, a system of classification for our Tasmanian igneous rocks founded upon varying acidity or basicity, it appeared far and away ahead of systems based upon structure, mineral contents, age, or geological occurrence. In that it rests upon a

chemical basis, it still appears to me to possess advantages over structural or mineralogical systems. Without committing my colleague to my present view, I may say that the advances in our science during recent years have led me to recognise deficiencies in the old scheme. I venture to think that it fails in assigning the proper values to chemical relationships and differences. By reason of the importance it attaches to the silica percentage, it severs related rocks like granite and syenite, and at the same time associates unrelated rocks like elaeolite-syenite and normal syenite. In nature, the former pair blend together often enough in the same mountain massif, the latter never, for they apparently represent distinct magmas, and their respective dyke-rocks and effusive equivalents have their own distinctive stamp and character. The dyke-rocks of elaeolitesyenite, for instance, will be tinguaites, not minettes, and its effusive product is phonolite, not normal trachyte. I look upon a system of classification as faulty if it neglects to take account of these important facts. In nature, too, I

may add that granite and diorite often behave in the same way, the one passing by easy transitions into the other in the same rock-mass. The United States Survey Officers have introduced the name granodiorite for this passage-rock. Syenite also graduates into diorite in a precisely similar manner. It would hence appear natural to group these three families of rocks in one great natural division. Rosenbusch, while adopting a totally different principle of classification, propounds a granito-dioritic magma, which corresponds with the division just suggested.

Having marked off the rocks of this magma, there remain only two other great divisions, which are easily appreciated, namely, elaeolitic-leucitic rocks and gabbro-peridotite-pyroxenite rocks. As we shall see later on, chemical considerations support this grouping.

Mr. Alfred Harker, in his interesting little book, "Petrology for Students," adopts plutonic, hypabyssal, and volcanic as the three primary divisions, and groups the rock families in these in order of increasing basicity. Thus, under plutonics he groups granites, syenites, diorites, gabbros, and peridotites; under hypabyssal rocks he ranges granite porphyries, ceratophyres, diabase, lamprophyres; under volcanics, rhyolites, trachytes, phonolites, andesites, basalts.

This is practically the principle followed by Rosenbusch, though the details are not the same, and the German author's dyke-rocks are differentiation products, not merely hypabyssal rocks. It is really the recognition of variety of geological occurrence, involving variety in structural characters. It has the disadvantage of bringing together in one assemblage rock families which are widely separated in constitution. Hence it is, I submit, unavailable for primary divisions.

The other great German petrographer, F. Zirkel, bases his primary divisions on mineral characters; viz. :—1. Rocks with dominant alkali felspar; 2. Rocks with dominant lime-soda and lime felspars; 3. Rocks without felspar. Under this grouping we have granitic and elaeolitic rocks thrown together in the same division.

The French school also classifies according to the felspars. Though the method is empirical, the results approximate to the truth, for the contents and proportions of alkalies are involved. For example, in syenite and granite potash felspars are the dominant ones; in diorite, Na2O has increased; hence, the felspar is soda-lime. In gabbro, alkalies are low and CaO high; hence, the felspar is lime-soda. In nepheline syenites, the excessive alkalies show themselves in all the characteristic minerals of that division.

The United States petrographers are in general agree ment as to a chemico-mineralogical basis of classification, and their groups do not differ from those already familiar to us. In 1897 a committee, comprising Professor Van Hise, J. S. Diller, W. H. Weed, H. W. Turner, and Whitman Cross, was appointed by the Director of the United States Geological Survey, to report upon the nomenclature and classification of eruptive rocks for adoption by the survey. The following scheme corresponds closely with the committee's recommendations::

An

The well-known groups are all here, and the recommendations indicate, as it were, the direction in which AngloSaxon petrographers may be brought into line. objection which may be urged, is that a certain want of proportion is apparent if these divisions are to be considered as primary ones.

I ought not to omit mention of a tentative classification of the rocks of New South Wales put forward in 1890 by Professor F. W. Edgeworth David. Its basis is the silica percentage, to which I have previously alluded. As a result, it groups together in one primary division such widelyseparated rocks as felspar and leucite basalts. With this principle of classification unnatural unions are inevitable.

I wish to be understood in this matter of SiO2 per cent. It is of distinct value in sub-divisions, and ought to receive expression in such, but per se it is not the all-powerful key which we perhaps have been accustomed to consider it. The more intimately we become acquainted with eruptive rocks, the more definite grows the conviction that no ingle element has governed the differentiation of their magma, not Si, nor Al, nor Na and K alone; their composition must be considered as a whole. If undue weight is laid on the SiO, proportion, the petrographer is sure to go wrong. Take a concrete example, one not far from home. Messrs. Guthrie, Woolnough, and Professor David have lately described a remarkable phonolitic rock from Kosciusko with a large proportion of nepheline, but with a low SiO2 per cent. (52-4 per cent.) The quantity of silica is lower than that of phonolite proper, and is, in fact, that of typical basalt; but to follow this as a guide would be hopelessly misleading. The authors say " The Kosciusko rock differs

?

conspicuously from typical phonolites in the following respects: -(1) low silica percentage; (2) entire absence of phenocrysts of sanidine. It is obviously a felspathoid rock, and although its silica percentage allies it with the basalts, its mineral constitution, chemical composition, and low specific gravity link it with the phonolites. As far as the authors are aware, it is unlike any rock that has hitherto been described from any part of the world."*

It is undoubtedly a rock of the foyaitic magma, and has no relation whatever with basalt.

Take, again, the new theralitic rock Euctolite, described by Rosenbusch from Umbriat. It is an olivine-meliliteleucite rock allied to madupite, which is a biotite-leucitediopside rock. Its SiO, per cent. is only 414, and would, if alone relied upon, indicate basaltic affinities, while its real relations are theralitic.

Among the signs of a growing discontent with the existing empirical schemes of classification, the critical contributions of Professor F. Loewinson-Lessing are important. His guiding motto is, chemical composition the basis of rock classification." His application of it is less acceptable. He uses a dualistic nomenclature, referring to the felspar and the colored constituents. Thus, with respect to Syenite, he distinguishes Orthoclase Syenite, Albite Syenite, Anorthoclase Syenite; and secondly, Hornblende Syenite, Augite Syenite, Mica Syenite, Aegirine Syenite, &c. Then, combining both constituents, he adopts such terms as Orthoclasehornblende Syenite, Anorthoclase-aegirine Syenite, &c. As will be seen later, I propose to avoid this dual terminology, by detaching the alkali granites and syenites from the normal series altogether. We shall then adopt the specific names which have been used for the different rocks of these groups, such as pulaskite, laurvikite, &c.

The plan which I now venture to propose is to arrange the primary divisions in accordance with the rock-magmas assumed by H. Rosenbusch, and we shall then obtain the series, as under:

1. The Granite-Syenite-Diorite series;

2. The Gabbro-Peridotite series;

The Elæolite or Nepheline series of rocks.

The sub-divisions, as they affect Tasmania, are represented in the attached scheme. For myself, I do not fear that this

Roy. Soc. N.S.W. Abstract of Proc., June 5, 1901.

Ueber Euktolith, ein neues Glied der theralithischen Effusivmagmen.

Vide Tschermak's mineralogische u, petrographische Mittheilungen, 1900.

scheme will not prove amply adequate to all requirements. I commend it to your criticism.

It is being continually urged upon petrologists by stratigraphical geologists that a nomenclature is needed which can be applied without resort to laboratory analysis. In short, field geologists wish to be able to name a rock at sight. Where it is impossible to determine the nature of rocks in the field, I do not see how a field-term can be anything more than a temporary expedient. The dyke-rocks and the marginally modified plutonics offer great difficulty, and I know of no Linnaean system which will extricate us. Field geologists must not complain if petrologists, before labelling a rock with a name, try by all the methods at their command to find out its mineral and chemical composition, the conditions of its consolidation, and its genetic relations. Nomenclature can then be discussed with some degree of confidence. Others wish to be able to name rocks without knowing their mode of occurrence, but F. Zirkel puts the case convincingly when he says that "petrology is a branch of geology, not of mineralogy, and its problems do not consist in determining specimens from unknown localities, nor are they solved in that manner. A rock specimen is not only an aggregate of such and such minerals, with a given structure and composition, but is also at the same time a piece of the earth's crust, which has played a definite geological rôle in the place whence it was taken; if that rôle be unknown, one essential element of a decisive diagnosis is lacking."*

In the sub-divisions I have followed the usual plan of giving to geological occurrence its legitimate expression under the terms plutonic, dyke and effusive. But these terms must not be used slavishly, or difficulties will be encountered. Moreover, the dyke-rocks are exceptional in that they are not mere intrusives, but have a chemical significance. Granitic, basaltic, diabasic dykes do not come within Rosenbusch's definition of dyke-rocks, though they have the dyke form. Dyke-rocks proper are differentiation products of the magmas of plutonic rocks. Naturally, each group of plutonics may be expected to be accompanied by its own dyke group. W. C. Brögger proposes the term complementary rocks" for types differentiated out of a common magma. +

Some of the dyke-rocks, e.g., lamprophyres in France and tinguaite porphyry in Brazil, have been shown to occur in

*Lehrbuch der Petrographie, F. Zirkel, 1893, vol. i., p. 840. On the basic eruptive rocks of Gran. Q.J. Geol. Soc., 1894,