Go to National Library of New Zealand Te Puna Mātauranga o Aotearoa
Volume 26, 1893
This text is also available in PDF
(3 MB) Opens in new window
– 368 –

Art. XLII.—Tridymite-Trachyte of Lyttelton.

[Read before the Philosophical Institute of Canterbury, 6th September, 1893.]


The volcanic system of Banks Peninsula presents to geologists an exceedingly interesting if somewhat complicated problem. Since the earliest days of colonisation it has been the subject of elaborate investigations by the officers of the Geological Survey of Canterbury, but no one has done more able and lasting work in this district than the late Sir Julius von Haast, formerly Provincial Geologist of Canterbury.

According to this observer—and his views have been accepted throughout the colony—the peninsula is composed of the ejecta and lava of a few large volcanic foci, the boundaries of which were determined by him, and laid down in the official geological map of Canterbury. Of these vents, the one situated in the depression now known as Lyttelton Harbour was, according to him, the largest and most important.

The harbour itself is about eight miles long by two broad, and is too large, according to Sir J. von Haast, to have been at any time the actual crater of a volcano, and he accounted for its dimensions in the following way:—

A large volcano, perhaps several thousand feet higher than the remaining caldera walls, occupied the site of the harbour, its eruptions being spasmodic and explosive. Between each two eruptions the vent is supposed to have been more or less choked up by congealed lava and agglomeratic accumulations. As, owing to the gradual diminution of the volcanic forces, the eruptions became less frequent the agglomeratic matter accumulated to a greater extent between two successive eruptions, and this, reacting again, caused the intervals to be still longer but the eruptions more energetic. After a long period of quiescence he supposes that the crater resembled that of Vesuvius before the eruption of 1813—a rocky plain, over which small ash-cones were built up—and that, as in the case of Vesuvius, the gases and steam generated in the volcanic laboratory beneath at length reached such a bulk, and had so great a tension, that they were able to overcome the resistance of the superincumbent matter, and a terrific eruption took place, blowing out a large quantity of rock, and leaving the hollow that now forms Lyttelton Harbour. Although it may

Picture icon

To illustrate Mr. P. Marshails paper on Tridyrnite Rocks.

– 369 –

seem bold to invoke the aid of such a vast explosion, we must remember that even in recent times there have been cataclysms that have equalled in intensity and magnitude that required for the formation of Lyttelton Harbour. The force expended at the Krakatoa eruption was quite as great as that required by hypothesis in the present case. Even in New Zealand we have had an illustration of the immense power exerted by imprisoned steam in the destruction of the terraces at Rotomahana, where a gigantic chasm was formed almost as large as Lyttelton Harbour.

Since this great spasmodic effort two other eruptions or periods of activity have been witnessed in this system. The first resulted in the formation of Mount Herbert in the former caldera wall, where the lavas which flowed down the sides of this secondary vent interfere with the symmetrical appearance of the walls.

By the last eruption Quail Island was formed. This eruption is unimportant; it does not seem to have been attended by any explosive action, and to have attained to only extremely small dimensions.

As would be expected from the spasmodic character of the earlier eruptions, a large number of radiating cracks were torn open in the surrounding crater-walls, and into these the magma was injected, giving rise to a well-defined system of dykes, which preserve an astonishingly constant direction, width, and composition over large horizontal and vertical distances. From a careful examination of these dykes it was ascertained that, with a few exceptions, they can be classed in two systems. Of these the most important radiates from a spot situated a little to the south of Quail Island, while the other proceeds from the centre of a shallow bay to the south-east of the former one.

The rocks of the Lyttelton system are, with a few exceptions, members of the basic series of igneous rocks, the commonest species being one that is best named olivine-andesite. Andesites without olivine have also been found, while basalts, especially in the later eruptions, are frequent, being, as a rule, very finely grained. Trachytes also occur, particularly in dykes, and from the prevalence of dykes of trachyte Sir Julius von Haast drew an important induction in support of Durocher's theory of volcanic action and the origin of the eruptive rocks. Rhyolites are also found as members of the very earliest eruptions, and still crop out at Governor's Bay, on the shore of the harbour.

The most anomalous of all the rocks occurs on the Lyttelton-Sumner Road, about half a mile from Lyttelton. This rock was discovered by Sir J. von Haast, and was called by him a domite. A chemical analysis was made by Professor Bickerton,

– 370 –

which is quoted on page 300 of Haast's “Geology of Westland and Canterbury.” Subsequently the rock was examined by Professor Ulrich, and he discovered the rare mineral tridymite in it. Specimens were also sent to Professor Von Rath, but no detailed description of the rock has yet been published, though a collection of volcanic rocks was described in Germany.

Professor Von Haast, in the “Geology of Canterbury and Westland,” describes this rock as “a remarkable trachyte rock, interstratified between two others of a basic character.” He thus evidently considers that it comes from the same source and has the same origin as the other rocks of the Lyttelton system. After a careful examination of the rock itself and the surrounding lavas, the writer has been forced to form an opinion directly opposed to the expressed and written ideas of the professor.

As it would be impossible to give an intelligible description without the aid of a diagram, part of an official chart of the harbour and an enlarged rough sketch-map of the area examined are added (Pl. XLIV.). In the chart it will be found at the point A. The chart has a scale of 1 ⅔in. to the mile. The sketch-map is intended to show only that part of the district that has been actually examined.

About half a mile from Lyttelton the Lyttelton-Sumner Road passes an abrupt wall of a whitish rock, K, about 30ft. high, extending about 70 yards up the face of the hill and a few yards below the road, the wall being nearly at right-angles to the direction of the road. As it is followed up the hill this wall gradually decreases in height, and is ultimately on a level with the surrounding ground. Higher than this the outcrop cannot be traced at this point, but it turns sharply to the right, and runs for some distance parallel with the road. The highest point of this outcrop is about 490ft. above sea-level. Between this height and 690ft. the slopes are thickly covered with grass, except in one or two places where basic lava that apparently overlies the trachyte crops out. At a level of 690ft. above sea-level there is another wall-like outcrop, which runs almost parallel to the direction of the road, and is fairly constant in elevation. To the right it gradually gets smaller, and disappears at C. On the left it ends somewhat abruptly at D. From this point to the top of the hill, 770ft., all the rock seems to be trachyte of the same character as the first outcrop, and its resemblance is borne out on microscopical examination. At E, along the crest of the hill, the trachyte seems to disappear, and a little further on a wall-like buttress of basic rock stretches across at right-angles to the axis of the hill. Descending the hill from E to F, the ground is strewn with boulders of the same rock; and at F, on the

– 371 –

shore of the harbour, there is a small cliff of the same rock, with several small sea-worn caves. The cliffs continue to bound the shore to the point H, whence it slopes up the hill to K, which was our starting-point.

There are four well-defined dykes in this rock, three of which, M and N, on the shore, are of small size, while the fourth, which cuts into the outcrop K, is 6ft. to 7ft. wide.

At the top of the hill there is clear evidence of a large dyke, which runs parallel to the length of the hill, along its summit. In breadth the dyke measures about 20ft., and in length about 200 or 300 yards, its bearing being E. 41° N.

It is evident that the trachyte lava issued from this dyke, as outcrops are found the whole way from the top to the bottom, and are generally of a vesicular character. If, as Sir J. von Haast suggests, it flowed from a central crater, situated near Quail Island, it would necessarily, like the other rocks of the harbour, present a single face of moderate breadth, and of a tolerably constant altitude above the sea-level.

It is possible that the lava flowed from a crater that occupied the position where the Town of Lyttelton now stands. This, however, is highly improbable, as no independent evidence exists of the activity of a vent situated there. No system of dykes has been discovered; none of the other hills enclosing the depression afford evidence of such a vent, while on one side all traces of crater-walls, if they ever existed, have been removed.

The very appearance, too, of the hill under consideration gives one the idea that its origin is not the same as that of the others, for, while its surface is, generally speaking, smooth and rounded, and resembles the slopes of Mount Herbert, the others, almost without exception, present that series of sharp, steep walls rising tier above tier that so plainly indicates to the geologist that they are formed of lava-flows lying one over the top of another with a moderate angle of inclination. These considerations, and the fact that a well-defined dyke exists at the summit, must remove all doubt as to whether an independent origin should be assigned to this small system of volcanic products.

As the age of the whole system of Banks Peninsula is not yet settled with any exactitude, it would be idle to attempt to ascertain the precise geological age at which this minor eruption took place. There is, however, little difficulty in ascertaining its age relatively to the other eruptions of this volcanic system.

It is stated by Sir J. von Haast that the original crater was the only one whose eruptions were of a sufficiently spasmodic character to rend fissures in the surrounding rocks, which, on being filled with igneous matter, form dykes.

– 372 –

Now, it is evident that this large dyke must have been formed during or preceding such a paroxysmal eruption. Its age, then, must be assigned to some period during which these eruptions were still in full force. Again, it is evident that the caldera of Lyttelton had already been formed when this eruption took place, for the lavas that were emitted from this dyke flowed down the face of the harbour-walls when they were approximately of their present shape and form, otherwise during the formation of the caldera this small accumulation would have been blown out of existence. Thus its age can be assigned as not earlier than the latest phase of eruption.

Lastly, the presence of other dykes piercing this rock proves that the convulsion by which the large dyke was formed was not the last effort of the declining volcano, but force still resided in it sufficient to crack the surrounding rock, and volcanic magma was still present in sufficient quantity and under sufficient pressure to fill these cracks up to the level of the surface, and thus form dykes.

The age, then, of the system must be stated as younger than the most violent paroxysms of the central volcano, but in all probability older than the Mount Herbert system, for these lavas are in no place seen to be pierced by dykes, and the eruption was therefore subsequent to the convulsions of the central crater.

The presence of sea-worn caves even near the top of the hill does not help us much, as it has been shown by Professor Hutton and others that within the Pliocene period great oscillations of level have taken place.

In macroscopic appearance the trachyte previously mentioned resembles a rhyolite in its very light colour, but no quartz crystals can be seen.

The colour is almost white in places, but generally iron-oxide has segregated in cracks owing to weathering, thus giving it a banded and sometimes almost spherulitic appearance.

Large crystals of plagioclase can be distinguished, the striation often being visible with a simple lens. The rock is generally vesicular, and it is in these vesicles that glass-clear tridymite crystals are seen, and frequently appear to have a hexagonal outline.

The texture is porphyritic, the phenocysts being invariably feldspars.

There are two well-developed divisional planes in the rock, one being parallel to its surface and the other parallel to the direction of flow, showing that the cooling proceeded from the surface as well as from the sides.

No macroscopical difference can be seen between the rock on the summit of the hill and that on the sea-level, except

– 373 –

that the former is, if anything, the more vesicular of the two.

The mineralogical structure of this rock presents many peculiarities, both in the nature of the minerals themselves and in their association with one another.

Tridymite occurs in considerable abundance, for, although thirty-six sections of the rock were cut, some crystals of the mineral occur in every section. The feldspars, both ortho-clastic and plagioclastic, frequently occur in large porphyritic crystals, and present the peculiar feature of a central core of plagioclase surrounded with a mantle of orthoclase or sanidine. Ferro-magnesian constituents are rare, or entirely absent, but magnetite occurs in considerable quantity, while needles of apatite penetrate the ground-mass and feldspars. As an accessory of somewhat doubtful occurrence, zircon may be mentioned.

Tridymite, although often present in the ground-mass, is generally found attached to the sides of vesicular spaces, and sometimes completely fills the smaller vesicles. The crystalline groups are, as a rule, of irregular shape, but in some a fairly regular hexagonal boundary may be observed. They are quite transparent, and possess a vitreous lustre.

In sections they appear generally as rounded aggregates, quite clear and transparent, with numerous cracks that resemble cleavage. With polarised light, however, they break up into a number of irregularly-shaped areas of extremely minute dimensions, but all possessing different optical orientation. To see the structure distinctly, a magnifying-power of 70 diameters or more should be employed, and it will then be noticed that, although irregular, there is an approach to the hexagonal boundary in the majority of the plates. Each of these areas undoubtedly represents a distinct individual, but, owing to their extremely small dimensions, it was found impossible to isolate any one of them and submit it to optical examination with the hope of forming any ideas as to the system of crystallization. The peculiar irregular structure of the aggregates is well shown by altering the focus of the microscope by means of the fine adjustment, when it will be seen that, even in the thinnest sections, there are several layers of crystalline plates.

The aggregates are frequently traversed by cracks which seem to bear no definite relation to the outline of the individual grains of the aggregates themselves. A peculiar feature of many of these grains is a radial structure (shown in Pl. XLVIII., fig. viii.). Although not universal, this structure occurs in the majority of the grains. Figs. vii. and viii. were both drawn from a section beneath a magnifying-power of 70.

– 374 –

Fig. vii. shows the normal structure of tridymite with polarised light, except that the different areas are not shaded as they appear beneath the microscope. The interference colours are always low, but it was found impossible to give the general effect by shading the different areas.

Feldspars.—These are perhaps the most interesting of all the minerals in this rock, for the sections serve to show that the isomorphism of orthoclase and the more acidic plagioclases (andesine) is almost exact—so nearly so that crystallization of sanidine can proceed with as great energy round a core of plagioclase as round one of sanidine.

Sanidine occurs in beautiful glassy porphyritic idiomorphic crystals, as well as in granular aggregates with irregular outlines. The porphyritic crystals are generally of small dimensions, and often give square sections, which shows that the crystals are not generally elongated in the direction of the chief axis. Cleavage is plainly seen in most of the sections, and a faint zonal structure can generally be observed. Between crossed nicols the interference colours are low, greys and bluish-grey of a low order being general, and affording a marked contrast with the brilliant interference colours of plagioclase in the same section.

Twinning is exceedingly common, and, though well-developed examples of the Carlsbad type are frequent, the majority of the twins are irregularly penetrating, frequently without the slightest indication of any regular law; but at other times twins resembling those formed on the Baveno and Manebacher law have been observed.

The Carlsbad twins—composition plane ∝ ρ ∝ (010), twinning plane ∝ ρ ∝ (100)—are numerous, occurring in crystals of all sizes, even the microlites in the base sometimes showing this type of structure. Some examples of these twins are given in the figures.

A twinned crystal that bears some resemblance to a Baveno twin is shown in Pl. XLVIII., fig. ix. Unfortunately, this crystal seems to have been considerably corroded before the rock cooled, and the margin has therefore become somewhat rounded. Supposing this to be a Baveno twin, it would consist of four individuals such as those figured in Dana's “Textbook of Mineralogy” (page 100, fig. 325, and page 325, fig. 587). The composition and twinning-plane are the clinodome 2 ρ ∝ (021). Although the traces of the twinning-planes are not exact diagonals, they do not show more irregularity than many of those of the Carlsbad twins. A very similar crystal has been found in another section; while in a third there is another, consisting of two individuals, the twinning trace being more nearly a diagonal.

Irregular penetrating twins are frequent, and some are

– 375 –

drawn in the annexed figures. They do not call for any special description.

In order to show clearly that the supposed sanidine is not plagioclastic feldspar cut in the direction of a plane parallel to the brachypinacoid ∝ ρ ∝ (010), Pl. XLVII., fig. v., may be mentioned. In the large crystal drawn in this section the traces of the faces which appear longest have cleavage-cracks parallel to them, and must therefore be the traces of the base O P (001) and the clinopinacoid ∝ ρ ∝ (010).

As these traces are almost exactly at right-angles to one another, and extinction takes place when the cross-wires are parallel to these traces, the section must be cut in the zone of the orthopinacoid and base. If, now, the mineral were plagioclase, twinning parallel to the brachypinacoid (corresponding to the clinopinacoid of sanidine, as shown below) would be observed in such a section—that is, supposing the usual isomorphic relations to hold good, as they will be shown to do when the plagioclases are considered. The fact that the mineral extinguishes parallel would in itself generally be considered sufficient to show that it is sanidine and not plagioclase. The truncations of the angles are, of course, due to the development of a clinodome. The same method of reasoning may be applied to fig. vi., Pl. XLVII., where there is a core of plagioclase surrounded by a broad rim of sanidine. This example is even more conclusive than the last, as here the plagioclase is twinned on the albite type, and it cannot therefore be pleaded that the mineral is untwinned plagioclase. Since sanidine is doubtless present in this crystal, it may be said with safety that the other unstriated feldspar possessing similar interference colours is sanidine.

Plagioclase is also present in crystals with idiomorphic outlines, and generally of a far larger size than the crystals of sanidine. Almost invariably, however, there is an investing mantle of sanidine, which sometimes is of far larger diameter than the core of plagioclase, but in other cases far smaller. In general, the plagioclase can hardly be called idiomorphic, as it passes in many cases gradually into the surrounding sanidine, the exact boundary-line being hard to determine. Occasionally, as in Pl. XLVI., fig. iv., there appears to be an outer zone of plagioclase possessing different optical orientation from the inner one. Cleavage-cracks can seldom be seen, and zonal structure is rare and inconspicuous when developed.

Twinning is splendidly developed, according to three well-defined laws—Carlsbad, albite, and pericline. The Carlsbad twins are exceedingly common, the two halves being frequently as sharply defined as in sanidine. The investment of sanidine, in every observed case but one, is also twinned, but its orientation is different from that of the plagioclase, although the

– 376 –

composition plane is a continuation of that of the plagioclase. Since these two planes are coincident, the plane of twinning on the Carlsbad type in the plagioclase must be parallel to the clinopinacoid in the sanidine. But the twinning-plane of the albite lamellæ is parallel to the twinning-plane of the Carlsbad twins, and, as, according to the albite law, the lamellæ are twinned parallel to the brachypinacoid, it is evident that the composition plane of the Carlsbad twins is also the brachypinacoid. Hence in these crystals the clinopinacoid of the sanidine corresponds with the brachypinacoid of the plagioclase. The Carlsbad twins in the plagioclase show the same irregular intergrowth that forms a noticeable feature in the sanidine.

Albite twinning is exceedingly common, the lamellæ varying greatly in width, but on the whole they are narrow, the width being constant throughout their length. In a few cases the lamellæ are curved or broken, and possess undulose extinction, thus showing that the rock was subjected to considerable pressure or tension previous to its extrusion.

If a section of plagioclase is cut at right-angles to the brachypinacoid the extinction of the adjacent lamellæ of the twins would make equal angles on each side of the cross-wires. Although in the sections none of the plagioclase crystals are cut precisely in this direction, some of the surfaces coincide approximately with such a plane.

The following have been measured:-

Extinction on Right Side of Cross-wires. Extinction on Left Side of Cross-wires.
1. 28° 1. 9°
2. 14° 2. 13°
3. 23° 3. 9°
4. 16° 4. 23°
5. 15° 5. 11·5°

These results, although rather high, tend to show that the species of plagioclase is andesine, a conclusion that will subsequently be shown identical with that derived by chemical analysis. It was found impossible to detach cleavage-flakes, so that the extinctions on the base or brachypinacoid could not be determined.

Twinning after the pericline law (twinning-plane the “rhombic section”) is not nearly so frequent nor so well developed, the lamellæ being as a rule of variable length, frequently not traversing the whole breadth of the crystal, and leaving untwinned feldspar between the adjacent twins.

Plate XLVIII., fig. x., shows a crystal where this type of twinning is seen in combination with twinning after the albite law. The drawing is a faithful representation of the crystal as

– 377 –

it appears between crossed nicols when the longer side of the crystal is inclined at an angle of 33° with the cross-wires. It would appear that h, e, d are broad lamellæ of pericline twins, n being a tongue of feldspar twinned on the albite law, the lamellæ appearing on rotating the stage. The lamellæ h, e, d are themselves striated with albite lamellæ, which also appear at e, c. The lamella d, on the other hand, has subsidiary striation due to twinning on the pericline law. c d, e f, all extinguish together, as do n h; mm are grains of magnetite. The whole is surrounded with a mantle of sanidine, a a.

Magnetite is present in every section, but varies considerably in abundance. The grains exhibit great variation in boundary, as the usual form of the mineral is the rhombic dodecahedron and octahedron. Some grains show a fairly regular hexagonal boundary, and it is possible that ilmenite, which crystallizes in the hexagonal system, is present.

Augite is of doubtful occurrence, for though between thirty and forty sections have been cut there is no single occurrence of this mineral which cannot be questioned. One section shows an opaque crystal with octagonal outline, which is probably a section of augite in which both prisms as well as pina-coids are developed. As, however, almost the whole of the crystal has been changed into iron-oxide, its optical properties cannot be investigated.

Apatite is present in numerous prismatic needles which pierce the feldspar and ground-mass, and was therefore the first mineral to crystallize out of the magma.

Zircon seems to be represented in one slide where there are two short prismatic crystals terminated by obtuse pyramids, and possessing straight extinction as well as strong double refraction, indicated by brilliant colouring between crossed nicols. It would, however, be rash to assert from these isolated examples that zircon occurs in the rock, especially as it has been found in no other slides, including those prepared by Professor Ulrich.

The ground-mass of the rock is composed of feldspathic microlites, tridymite, and magnetite, and in rare instances there are globules of a greenish glass. Beneath, the quarter-inch objective it cannot be determined to what species of feldspar these microlites belong, but, since sanidine has in the porphyritic crystals evidently crystallized subsequently to the plagioclase, we have a certain amount of right to infer that the microlites consist chiefly of sanidine. Irregularly-bounded tablets of tridymite will be observed in very thin sections to occupy a large portion of the ground-mass, but in thicker sections they are frequently over- or underlaid by microlites. It will be shown afterwards that a large proportion of the base must consist of tridymite.

– 378 –

Iron-oxides (magnetite, and sometimes hæmatite) are frequent in small specks without crystalline boundaries. Hæmatite, with a little limonite, is particularly abundant in cracks, and in all probability results from the further oxidation and hydration of the magnetite.

Beneath the highest power of the microscope (700 diameters) no isotropic matter indicating the presence of interstitial glass can be seen, but occasionally, especially from the vesicular upper surface of the lava-flow, there are inclusions of a greenish glass in globules of a spheroidal form, the structure in one or two cases being almost pisolitic.

A few of the sections were examined in convergent polarised light, and the following results were obtained:—

The large crystal in Pl. XLVII., fig. v., shows two hyperbolic brushes which meet almost in the centre of the field, and the section must therefore be cut almost exactly at right-angles to an axis of elasticity. Since in such a thin section both hyperbolas appear, and when furthest apart they barely disappear from the field, the axis of elasticity must be the acute bisectrix. With the quartz wedge the hyperbolas are far more widely separated, and the result therefore is equivalent to the thinning of the section, which shows that the optical signs of quartz and sanidine are opposite, and since quartz is positive the acute bisectrix in sanidine is negative. Further, since the section was shown to be cut in the orthopinacoid-basal zone, the acute bisectrix must be at right-angles to the orthodiagonal.

The acute bisectrix is thus shown to be negative, and at right-angles to the orthodiagonal, a result in accordance with the general optical properties of sanidine or orthoclase (Dana's “Text-book of Mineralogy,” p. 325).

Plate XLVII., fig. vi., in convergent polarised light only shows one hyperbola, and is therefore not cut at right-angles to an axis of elasticity, although it is cut perpendicularly to the plane of the optic axes.

In convergent polarised light tridymite only shows faint indications of hyperbolic brushes, which appear only in the edges of the field.

Plate XLVIII., fig. ix., the supposed Baveno twin, gives uniform phenomena all over the crystal as far as can be seen, but, as it is not cut at right-angles to an axis of elasticity, this is not certain, as the brushes merely sweep across the section apparently in the same direction in all parts of the crystal. This militates considerably against the supposition that the crystal is a Baveno twin.

Chemically, this rock presents rather peculiar features, as will be seen from the following analysis. The first two analyses are of the same specimen, while the other

– 379 –

three are of another specimen, collected from an adjacent locality:—

Specific gravity, 2·351—2·415.
SiO2 72·32 73·07 71·57 71·20 71·09
Al2O3 13·95 13·75 16·07 15·57 15·45
Fe2O3 2·44 2·55 1·51 1·49 1·50
FeO 0·28 0·28 0·34
MnO Trace Trace Trace Trace Trace
CaO 3·41 3·27 3·52 3·46 3·25
MgO 0·88 0·99 0·90 0·87 0·89
P2O5 Trace Trace
K2O 2·46 2·46 2·35 2·35 2·35
Na2O 4·60 4·60 4·81 4·81 4·81
H2O, and loss 0·07 0·07 0·07
100·06 100·69 101·08 100·10 99·75

The large percentage of silica combined with such, a considerable percentage of lime is a very unusual feature, but would be expected when the large amount of plagioclase is taken into consideration. The excess of silica over and above that required for combination with the bases is of course present in the mineral tridymite. The analyses tend to show that the feldspar present is chiefly a species of plagioclase—i.e., a soda-lime variety—and this is confirmed by optical examination.

The presence of manganese was indicated by the bluish-green colour of the fusion mass, but this was not strong enough to warrant a special quantitative determination.

The percentage of magnesia shows that some ferro-magnesian mineral is present in small quantity, but microscopical examination shows that they are practically absent. Some of the larger accumulations of iron-oxide may possibly indicate the former presence of some ferro-magnesian mineral which has been re-fused owing to the relief of pressure, or some other changed condition subsequent to its original crystallization.

Phosphoric acid is of course accounted for by the presence of apatite crystals. No test was made for this substance in the last-three analyses.

In the first two analyses the iron was all converted to the ferric condition before precipitation, and no determination of ferrous oxide was made.

From these analyses a rough approximation may be made as to the amount of free silica present in the form of tridymite, for, assuming that all the potash is in combination with silica in sanidine, and all the soda goes to make up the plagioclase (andesine), the excess of silica over that required

– 380 –

for these combinations will be present as tridymite, as no interstitial glass has been detected.

Taking the average composition of sanidine as that given in Dana's “Text-book of Mineralogy,” we find that it contains 16·9 per cent. K2O and 64·7 per cent. SiO2. In a rock containing 2·46 per cent. K2O, 9·42 per cent. SiO2 would be required for combination. Again, CaO and Na2O make up 14·7 per cent, of andesine, while the percentage of silica is 59·8. Assuming that Na2O and CaO are to a limited extent mutually replaceable in the andesine molecule, and taking the molecular weight of Na2 (46) as equivalent to that of CaO (40), we have in this rock 8·06 per cent, of these bases, and they would require 32·60 per cent, of silica for complete saturation. Altogether, then, 42·02 per cent, of silica is required for combination, and the remaining 29 per cent, will be represented by tridymite. As, however, a very small proportion of the rock as seen in section is tridymite, it is fair to assume that a considerable amount must exist in the ground-mass.

As tridymite is stated to be soluble in caustic soda (Rosenbusch: “Microscopical Physiography of Rock-forming Minerals,” translated by Iddings, p. 174), an attempt was made to estimate its percentage by taking advantage of that fact: 29·95 per cent. SiO2 was dissolved out of the rock, but, as 15·05 per cent. Al2O3 was also present in the solution, the result is not satisfactory.

An analysis was also made of the feldspar, but it was found extremely difficult to isolate an appreciable quantity, and the result of the analysis cannot be considered strictly accurate:—

SiO2 60·56
Al2O3 22·05
CaO 7·25
K2O 3·54
Na2O 7·93

The result is almost the same as would be expected on consideration of the bulk analysis of the rock, except that the percentage of K2O is rather high. Sanidine forms only a small border round the relatively large crystals of plagioclase that were the only ones obtainable for analysis.

No attempt was made to determine the species of plagioclase by means of its specific gravity, as the difference in this physical quantity between two closely-allied species of plagioclase is extremely small, and would be completely masked by the lower specific gravity of the mantle of sanidine. An approximate correction for this would give an eminently unsatisfactory and wholly unreliable result. The same applies to Szabo's flame reactions.

– 381 –

Although fully recognising the danger of theorising upon any subject when the data are insufficient, I cannot refrain from offering a suggestion as to the cause of silica crystallizing in the form of tridymite. The available data to reason upon are—

  • 1. The invariable presence of tridymite in cavities;

  • 2. Its frequent radial structure;

  • 3. Its non-inclusion in other crystals.

All these point to the conclusion that the mineral was deposited subsequently to all the other minerals of the rock, and from a solution in some solvent. Silica if uncombined is well known to be insoluble in pure water at atmospheric pressure; but there seems to be every reason to suppose that it is soluble in water above its critical temperature and under considerable pressure. The experiments of Daubrée with water in glass tubes confirm this, and the theory to account for the fact that quartz is the last mineral to crystallize in granite shows that it is generally believed that silica can be dissolved in water of a sufficiently-high temperature and under great pressure. In lava, prior to its extrusion, the water present would probably be capable of dissolving silica, but on the eruption of the lava the pressure would be suddenly relieved and the temperature quickly lowered, and consequently the silica would be precipitated. Owing to this rapid precipitation the silica molecules might be unable to arrange themselves in their ordinary arrangement, or the molecular forces, owing to direct solution or the subjection to heat, might well be different from the ordinary molecular forces, and would tend to form a different geometrical solid from that formed under the normal conditions.

That the molecular forces under certain conditions are liable to variation is well shown by the production of dimorphous forms in Ca CO3, native sulphur, and other minerals; and there seems to be no physical reason why variation should not exist in the case of others when the conditions accompanying crystallization also vary.

The extreme minuteness of the individual crystals would seem to indicate rapid precipitation, while inclusions of steam or other fluids would easily find their escape through the divisional planes between the different crystal plates.

It does not seem probable that high temperature alone, or sudden cooling, determines the system in which silica crystallizes, since in rhyolites and other lavas in which these conditions obtained the silica crystallized in the form of quartz. In the case of holocrystalline rocks, on the other hand, the cooling was extremely slow; and in the case of older lavas the molecules may have had time to readjust themselves in accordance with the normal forces after extrusion. In recent lavas

– 382 –

containing quartz the absence of tridymite may be explained on the hypothesis that steam was not present in sufficient quantity or of a sufficiently high temperature or tension to dissolve the silica.

The juxtaposition of sanidine and plagioclase can easily be explained by assuming that under like conditions the molecules of plagioclase would have greater tendency to accumulate than those of sanidine, and the plagioclase would accordingly tend to crystallize first. With slightly-changed conditions sanidine might also crystallize, and because of the almost complete isomorphism existing between it and plagioclase it would crystallize round the already-formed core of plagioclase.

Possibly there is a gradual increasing acidity in the plagioclase from centre to margin, as well as a gradual change from the base Na2O to K2O. This would better explain the presence of the mantle of sanidine, for during crystallization there would not only be a gradual increase in acidity in the successive layers, but also a corresponding gradual change in crystalline form. The centre of the core of plagioclase might then be very basic, while the outside rim might be acid sanidine. The occasional presence of an intermediate differently-oriented layer of plagioclase between the central core and the sanidine lends this view considerable support. Again, the boundaries of the plagioclase very seldom show that sharp definition that would be expected if the change was sudden; and frequently there appears to be no actual line where the plagioclase ends, but it appears to pass laterally into the sanidine.

The nomenclature of the rock presents considerable difficulty, for it seems to form a connecting-link between three classes of lava.

If classification according to percentage of silica is adopted, the rock must be classed with the rhyolites or liparites, though the presence of quartz crystals is generally adopted as a specific character.

If Rosenbusch's classification is adopted, we should hesitate as to whether the rock should be called a trachyte or an andesite, the distinction hinging mainly on the preponderance of sanidine over plagioclase, or vice versâ; and as, according to analysis, the latter appears to be in excess, the rock would have to be classed with the andesites.

According to its mineralogical composition, a place might be assigned to it among the dacites; but, at the same time, tridymite would have to be considered a mineralogical equivalent of quartz. The absence of any ferro-magnesian constituent must, however, tell largely against its inclusion in this class.

– 383 –

The name tridymite-dacite would give the idea of presence of free silica, as well as the preponderance of plagioclase over sanidine, but would imply the presence of augite or hornblende. On the other hand, the feel of the rock, its specific gravity, the absence of a glassy base, and its very light colour, as well as, to a certain extent, its highly acid composition, all tell largely in favour of its being pronounced a trachyte; and I have therefore preferred to call it an andesine-tridymite-trachyte, the two pronomens being used to indicate its exceptional mineralogical characters.

In order to demonstrate the relations of the neighbouring rocks to this trachyte, several sections have been cut and some analyses made of the lavas that appear to have, proceeded from the same orifice, as well as of the intrusive dykes and other neighbouring rocks.

Immediately underlying the trachyte, and undoubtedly proceeding from the same vent, is a black lava that resembles pitchstone except for large porphyritic crystals of feldspar. Cavities are fairly numerous, but not so abundant as in the overlying rock. Sp. gr. 2·46.

Under the microscope there are large idiomorphic crystals of sanidine, and occasionally plagioclase surrounded by sanidine, as in the overlying rock. The sanidine is generally twinned on the Carlsbad type, some splendid crystals of this twin being obtained. The plagioclase is twinned after both the albite and Carlsbad laws, but does not present such interesting or marked combinations as in the overlying rock.

Augite of a bright-green colour occurs in idiomorphic crystals, generally of small size, as well as in microlites. It is slightly dichroic, but calls for no special mention.

Magnetite and apatite are present in fair abundance. The ground - mass consists of microlites of augite and plagioclase and abundant interstitial brown glass. A quantitative analysis gave the following result:-

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

SiO2 67·63 67·36
Al2O3 18·86 18·90
Fe2O3 3·37 3·46
CaO 3·23 3·30
MgO 0·91 0·98
K2O 3·78 3·78
Na2O 4·66 4·66
102·44 103·44

The rock appears to be a dark trachyte, with a large percentage of plagioclase, and, from the close resemblance that the analysis bears to that of the overlying rock, there can be no doubt that they both have a common origin. The geologi-

– 384 –

cal relations as mentioned above also point to the same conclusion.

At the top of the hill along the strike of the dyke another dark-coloured rock occurs of a far more basic character. The geological relations of this rock to the tridymite-trachyte are, however, exceedingly difficult to determine, as the whole surface of the hill near the outcrop is covered with grass.

The rock contains porphyritic crystals of plagioclase twinned on the albite law, generally with idiomorphic outlines. A few crystals of sanidine were also seen in section.

Augite is abundant, brown in colour, and displaying no dichroism. Polysynthetic twinning parallel to the orthopinacoid exists in the centre of some of the crystals, the more peripheral portions being untwinned. No olivine was observed in section. The ground-mass is glassy, with feldspar and augite microlites, and abundant grains of magnetite.

An analysis gave the following percentages:–

SiO2 49·49
Al2O3 18·05
Fe2O3 12·94
CaO 9·81
MgO 3·24
K2O 1·30
Na2O 6·04

Both mineralogically and chemically this rock shows great resemblance to typical basalts, and should be classed with this species.

Sections were also made of the dyke that penetrates the rock at B. Feldspar of a plagioclastic variety is abundant, but no sanidine was observed. Augite is present in idiomorphic crystals of a brown colour, and is not dichroic. Apatite is particularly abundant, and grains of magnetite are of frequent occurrence.

SiO2 50·82
Al2O3 14·79
Fe2O3 7·99
CaO 11·10
MgO 4·12

The alkalies were not determined. The rock is undoubtedly one of the ordinary augite-andesites of the Lyttelton system, and bears no relation to the previously-mentioned rocks.

Another rock outcropping close to the tridymite-trachyte was also examined. Its geological relations could not be made out, as all the surrounding slopes are covered with grass.

Under the microscope numerous large porphyritic crystals of plagioclase can be seen, and a few of sanidine, but the two do not occur in contact.

Picture icon

Tridymite Rocks

Picture icon

Tridymite Rocks

Picture icon

Tridymite Rocks

Picture icon

Tridymite Rocks

– 385 –

No augite was seen, but a highly-dichroic mica with parallel extinction, probably belonging to the biotite series, was present in a few crystals.

The ground-mass consists of crowds of interlacing crystals showing parallel extinction, their length being many times as great as their width. Their polarisation colours are greys and bluish-greys, not so high as those of the feldspars. They would appear to be orthorhombic zeolites, though no fibrous aggregates have been detected.

Microscopically, the rock is of a greenish-grey colour, with shining lustre in places. The specific gravity is 2·65. A quantitative analysis resulted as follows:—

SiO2 58·93
Al2O3 23·95
Fe2O3 5·43
CaO 1·75
MgO 0·96
K2O 4·32
Na2O 5·61
H2O, and loss 1·36

These percentages bear a certain resemblance to those in phonolites, the high percentage of alkalies and small percentage of lime being particularly characteristic. As, however, the rock is evidently considerably decomposed, and no specimens of the unaltered rock could be obtained, little reliance could be placed on the analysis.

On digestion in cold hydrochloric acid a large quantity of gelatinous silica separated out, a reaction eminently characteristic of phonolites. It is highly probable, however, that the gelatinous silica is in this instance a product of the decomposition of the zeolites.

Judging from its chemical composition, it would appear highly improbable that this rock has a common origin with the tridymite-trachyte. No other rocks were found in the neighbourhood of the tridymite-trachyte except a few small dykes 6in. to 18in. wide, and, as they obviously have no bearing upon the origin of the lava-flows, they were not submitted to a critical examination.

In considering this small subsidiary eruption of the Lyttelton system as a whole, it may be said that the order of the extrusion of the rocks is wholly in accordance with Durocher's law of succession of igneous rocks. First we have an intermediate lava represented by the black trachyte; next comes the tridymite-trachyte, as a representative of the acid group; and finally the basic rocks, as represented by the basalt.

– 386 –

Although it is tempting to generalise as above, it must be carefully borne in mind that only three lavas have been found extruded from this vent, and there is little doubt that a more detailed examination would reveal the existence of other lavas, while there may also have been several previous eruptions, the lavas being covered by later accumulations.

It would therefore be safe to say, of all the rocks of this system so far examined, that the order of succession seems to be in accordance with that demanded by Durocher's theory.

Summary.—The tridymite - bearing rock mentioned in Haast's “Geology of Westland and Canterbury” was erupted from a dyke formed during the paroxysmal convulsions of the central crater after its actual activity had ceased. It is not interstratified with the other volcanic rocks of the Lyttelton system. After the lava-stream had been formed fissures were torn open in it by continued paroxysms of the central volcano, and magma was forced into them, thus forming dykes through the consolidated lava.

The absolute age of the rock cannot be determined, but its age relatively to the products of the large crater can be easily ascertained. Investigation of its mineralogical and chemical constitution shows that the rock should be classed with the trachytes, its special characteristics being denoted by the name andesine-tridymite-trachyte. The rocks that have been extruded from the vent appear to be in the order of succession demanded by Durocher's theory.

Explanation of Plates XLIV.-XLVIII.

Plate XLIV.

Fig. 1. Chart of part of Lyttelton Harbour.

Fig. 2. Enlargement of part marked A in fig. 1: OE, dyke at the top of the hill; R, Lyttelton—Sumner Road; K, first outcrop.

Plate XLV.

Fig. i. is a section showing the general structure of the rock. The magnifying power is 30 diameters. a a1 are crystals of tridymite, the characteristic structure being shown more clearly than it actually appears under the microscope; b is sanidine, with almost rectangular boundaries, to which the cleavage is parallel; c is an irregular segregation of crystals of sanidine, the boundaries between the parts possessing different orientation being indicated by lines—none of the grains show any striation, and the boundaries are irregular; d and e are grains of iron-one, probably magnetite, but possibly ilmenite; f is a vein of hæmatite or limonite that has evidently been formed during the weathering of the rock.

Fig. ii. is also a general section magnified the same number of diameters as fig. i. a and b are grains of tridymite; c is a feldspar crystal consisting of an irregular intergrowth of two crystals, sanidine and plagioclase, the former in both cases surrounding the latter; the large crystal is an irregular Carlsbad twin; f is a compound crystal of sanidine similar to c in fig. i; e, d, and k are other crystals of sanidine; h is a large grain of magnetite.

– 387 –
Plate XLVI.

Fig. iii. is a section in which tridymite occurs to a greater extent than usual. a, b, and c are grains of this mineral; e and d are untwinned sanidine; f is a grain of magnetite. In all of these sections there is no attempt made to give the exact appearance of the ground-mass, since beneath such a low power it merely presents a cloudy appearance.

Fig. iv. shows a large crystal of plagioclase surrounded by a ring of sanidine, with a small layer of differently oriented feldspar between. a is the sanidine, b the intermediate layer of plagioclase, and c the central core of plagioclase; a is extinguished in polarised light when the trace of the clinopinacoid makes an angle of 26° with the cross-wires; b is extinguished when the angle is 11·5°, and c when the angle is 33°. The colouring shown in the drawing is seen when the trace of the clinopinacoid makes an angle of 64° with the cross-wires.

Plate XLVII.

Fig. v. a is a crystal of sanidine which was described when treating of the mineral generally; b is a vesicle; c is another sanidine crystal.

Fig. vi. This also was described previously. a is the outer ring of sanidine; b is the central portion of plagioclase; c is a mass of iron-ore; e is tridymite; f, air-space, or vesicle; while d is sanidine that has apparently intergrown with the larger crystal, but has independent orientation.


Fig. vii. is a crystal of tridymite as seen under a magnifying-power of 70 diameters. The outlines of the different plates are quite noticeable with this power.

Fig. viii. is another tridymite grain in which an attempt is made to show the radial structure that is often so noticeable with polarised light.

Fig. ix. is the supposed Baveno twin, also magnified 70 diameters. Two opposite quarters of this crystal extinguish when the line joining a and b makes an angle of 11° with the cross-wires. The other two extinguish when the angle is 18°. When the angle is increased to 23° they are indistinguishable, and remain so until the angle is 79°

Fig. x. was previously mentioned as a combination of albite and pericline twinning, a a is the surrounding sanidine; b b are lamellæ, apparently of albite twins on a broad pericline face; c is twinned on the albite and d on the pericline law; e and f are also pericline lamellæ, twinned according to the albite law; h is another pericline plate, which in certain positions shows secondary twinning; n appears to be an albite lamella which shows no secondary twinning; m m are magnetite grains. The appearance shown in the figure is presented when the lines of albite twinning make an angle of 27æ with the cross-wires.

Fig. xi. is a compound grain of sanidine and plagioclase. a a a, &c, are sanidine grains, all with different orientation; b b are plagioclase; and c c are grains of magnetite. This kind of combination is very common in the rock. The last two figures are magnified 30 diameters.

All the crystalline outlines were drawn with the camera lucida, and, when shaded, polarised light was used.