The article here is а direct translation from the original Russian. Version. Adapted variant was printed in Proc. Univ. Bristol Spel. Soc., 1996, 20(3)


Switch to the Russian version


Vladimir A. Maltsev


Some considerations on cave minerals onthogeny: sulfates filamentary crystals and their aggregates

ABSTRACT. The article, based on observations from the Cupp-Coutunn cave system, criticizes genetical models for cave sulfates filamentary crystals, based on solutions seeping through porous substrates under some external pressure. The physical models of such seeping are built, and the seepage hypotheses is found to be inconsistent for natural environments. The main genetic mechanism is proved to be only local feeding, based exclusively on capillary pressure together with short-period (seasonal) humidity cycle. In some cases also the major role of sulfatereduction processes is found. Some new aggregates, based on filamentary crystals growth, are described.


Filamentary and acicular gypsum crystals (in some cases formed with other minerals), and their aggregates (antholites or flowers, "hair", "beards", "cotton", etc.), are rather common in caves, and are mostly studied well. Three main papers on cave filamentary crystals [6,7,13] suggest mechanisms for filamentary crystals growth in near-surface zone of porous substrates both for the cases of oversaturated and undersaturated solutions. With this, feeding mechanisms, described as seepage through the substrate under some external pressure, seem to be wrong for the most of aggregates, studied by author.

In this article we'll discuss genetic mechanisms for the filamentary crystals aggregates, majorly conflicting with the seepage feeding model, along with some new aggregates types. All of them are generated on the phase boundary between porous carbonate substrates, and the air. All the factografic material used comes from the Cupp-Coutunn cave system, Kugitangtou mountains, Turkmenistan, but just the same aggregates are found in a lot of caves [4].

Crystallization in the surface zone of solid substrates.

A typical gypsum antholite from Cupp-Coutunn, according to Malishevsky [7], consists from filamentary crystals 0.002-0.1 mm in diameter, if to sort out crystals, specific for clay substrates, that are thicker. "Hair" and "spiderwork" consist of even thinner crystals. This means crystallization in pores of corresponding diameter. Let's understand physics of seeping through such substrates. If we consider some permanent flow (appearing when feeding is external), we can meet only two "clear" situations, and their mixtures.

The first situation. The water level in the substrate is far enough from the surface to have "full-profile" solution meniscuses in the pores. Knowing the pores diametres , the surface tension of the water, contacting with vapor or air , and the wetting angle , we can estimate the capillary pressure range in the pores:

(for hair - even more),

with about 25% error, caused by local features of the solution and the substrate. This means, that in this case, if we want to have seeping through the substrate as a result of external pressure, we must have this pressure higher, than the capillary pressure - the last here appears as some hydrodynamic analog of friction at rest. As the capillary pressure reaches 1440 at., and having such hydrostatic pressures in a karst cave is impossible, we'll stop the further examination of this situation.

The second situation. When the substrate is completely water-filled (meniscuses are absent), we can have external reasons only as a source of excessive pressure - the capillary pressure is self-compensated. It's easier here not to explicit some general physics, but to take a typical example, and carry out calculations just for it. The result will be precise enough for the most of typical observations. Gypsum antholites in the Dvukhetazhnyi chamber grow on a roof of a small dead-ended passage, having another passage in 1m above. Let's calculate the expenditure of solution in pores. Here we can imagine only some river in the upper passage, having the lower passage dry (it seems impossible, but in the caves of Kugitangtou it's rather common). So, we have the pores length , the excessive pressure . The filamentary crystals have their diameters about here. The water viscosity for 20╟C is . Now we've got everything to apply the Poiseuille's equation:


Calculating the equation for our data, we receive the water expenditure in a pore about , that corresponds to the linear seeping speed about 40 cm/year (this is the maximum estimation, because of that the pores geometry is more complicated, than a cylinder, modelled in the equation). Considering the solution as saturated (other way we'll have crystallization outside the substrate), we can re-calculate it into the maximum possible gypsum growth speed - 0.4 mm/year. This estimation (if understood as maximum), doesn't contradict to other observations - antholites size comparison to the possible length of stabil microclimate periods, estimated through observations on fast-growing speleothems sizes and growth speeds - the estimations differ not more, than 3-4 times. But one conflict still remains. The modelling, presented above, is true for a single pore, or for substrates with equal-sized pores, impossible in a real cave. Pores dimensions always have the log normal distribution, and the limestone always has fractures. It's absolutely impossible not to have on the antholites growth area (several square metres) a lot of pores and fractures, having their diameter 2 and more orders larger, than the mean value. From the same Poiseuille's equation we receive, that the water expenditure is proportional to the 4-th order of the pore diameter, and the linear flow speed - to the 2-nd order of the diameter. If to consider the mean speed estimated together with the idea of existence of 2-order larger pores, we can see, that there is no theoretical possibility of complete evaporation on these pores openings. We must have excessive flows from them, and stalactites, growing from these flows. And stalactites are never singenetic to antholites. More, the generic view of the situation received, strongly contradicts the statistics of observations, showing antholites growth only in very dry cave areas [4,13], free from any water flows.

Photo 1. Gypsum antholites up to 3cm long, growing on an isolated limestone block 15cm large. Cupp-Coutunn cave, Nadezhda chamber. Photo: V.Maltsev.

It's easy to see from these estimations, that any mixed case, derived from these two situations, has enough reserve of irreality to be impossible. So, we can make a conclusion, that generally any external feeding hypotheses for the sulfates filamentary crystals growth on solid substrates can't be accepted. We must also understand, that this doesn't contradict the classical Maleev's book [6]. In the chapter on experiments it's explicited, that the filamentary crystals growth is received only with optimum pressure, combined with optimum pores dimensions. The last also means high regularity of the substrate, possible only for artificial substrates, used by Maleev. So, Maleev's mechanisms really work, but for real cave conditions, with less regular substrates, we must discuss some other feeding mechanisms.

If external feeding is impossible, than we must speak about local feeding mechanisms. This means, that all the moistening of the substrate, and all the seeping through it, are caused by capillary pressure itself. In this case it functions not as an obstacle, but as the main working force. There are again two such mechanisms possible. The first is a seeping through mechanism, when one side of the substrate contacts with some solution reservoir, and the other side appears in a dry place. In this case evaporation from one side makes the solution inside the substrate move. This mechanism doesn't contradict any theory, but if it works, we must have filamentary crystals growth on any bank of any permanent pool with near-saturated gypsum solution. All the observation statistics, both by author and from literature [4,13], shows, that filamentary crystals are never found near any pools. Maybe, dry enough conditions are simply impossible near water reservoirs, maybe something else - but it doesn't work. The second possible mechanism is the mechanism, when the substrate functions like a buffer, moistened and dried during some short-period humidity changing process, like the seasonal process, described in [8]. Of course, this mechanism also may include seeping, because it's possible to have main condensation from one side of substrate, and main evaporation - from some other side. Anyway, the antholites crust texture will have it's symmetry, controlled with evaporation/condensation physics.

Photo 2. Gypsum beard, 12cm long. Cupp-Coutunn cave, Pautinny maze. Photo: V.Maltsev.

As it was said by somebody wise, after all the possibilities except one, are excluded, the rest one is truth even if it seems improbable. So, let's examine closely the one variant left - seasonal moistening and drying process, caused by seasonal cave wind inversions. The boundary conditions are simple and real:

a) moistening can't be complete - the maximum water level must stay deep enough in pores, other way the aggregates seasonal dissolution or re-crystallization are inevitable. This means, that the solution capacity of the substrate must exceed the full-year condensation on it's surface, and the full-year condensation must not exceed the full-year evaporation. There are no doubts for possibility of such conditions in some caves;

b) the substrate must contain enough gypsum or other sulfate. This condition is less trivial, and for the firs look contradicts a lot of observations. For example, the limestone blocks in a large collapse of the Nadezhda chamber, are covered with gypsum antholites from all sides. If we see such block, 15 cm large, having on it's surface regular covering of antholites 2-4 cm long and 3-6 mm thick (photo 1), and the block itself contacts surrounding blocks only in 4 points, each about 0.5 cm2 large, we see no evident feeding possibility. If the feeding comes from surrounding blocks, the covering must have visible disymmetry. If the feeding is strictly local, there is no any possibility of having such quantities of sulfate neither as solution in the pores, nor in limestone itself. With this, the variant of periodical flooding is also excluded - we can see no traces of dissolution or re-crystallization, necessary for such soluble mineral, as gypsum. The only possible explanation - permanent gypsum generation inside the blocks. Bacterial sulfatereduction and sulfuroxidizing are known in Cupp-Coutunn [5]. Existence of gas in the air and acid on limestone surfaces are usual. In moist periods both penetrate into the pores with the condensing water, and both take part in gypsum generation. This process becomes evident in places, similar to Nadezhda chamber, but is important in all the Cupp-Coutunn system - there is a strong correlation between intensity of sulfuric corrosion, and intensity of gypsum and epsomite filamentary crystals aggregates growth. We can also note, that gypsum antholites finds in other limestone caves are also often reported together with sulfatereduction.

There also exist some additional evidences of the short-period cyclic crystallization. The antholites crust texture has specific double-directional disymmetry, showing such cyclicity. The antholites are mostly developed on substrate ledges (where evaporation is forced), inside larger grooves - niches, fractures, under blocks (where condensation is forced). Absence of other disymmetries (for example, absence of fracture-controlled antholite lines), also shows absence of external pressure. Only in this case the growth speed in a pore is proportional to the first order of the pore diameter. It corresponds to observed splitting grades for antholites, and direct lengths disperse observations for beards and hair - they are all within the first order statistical models.

It seems necessary do discuss closely the most critical point of the suggested model - why, if we speak about a condensation/evaporation cycle, we can't find any traces of filamentary crystals aggregates corrosion directly by condensing water. In reality, such corrosion sometimes exist, but only for antholites, and very weak. In the case of poorly connected aggregates - beards, hair, spiderwork (photo 2) - it's absent completely, other way they are to be destroyed every year. There are three main reasons for it, listed in order of their importance increase:

a) the substrate surface curvity is generally less, than the aggregates surface curvity, thus setting the evaporation/condensation balance for the aggregates greater, than for the substrate;

b) condensation is not necessary caused directly by air overmoistening. More common immediate reason is in temperature jumps, always going together with the humidity jumps in the seasonal cycle. In this case a lonely filamentary crystal has very low thermal capacity, thus equalizing its temperature to air's quickly (the thermal capacity is proportional to the 2-nd order of the diameter, the heat exchange intensity and the condensation area - both to the 1-st order). So, the condensation per surface area unit on the substrate and on the aggregate differs at least 1 order;

c) rapid sucking in of the condensing water. The capillary pressure, estimated above, now all works for immediate sucking the water into the pores (except 1-2 molecules thick film, having very low mass-capacity). It's strong enough to leave no time for corrosion in the near-surface zone, especially just near the pores openings, where the crystals that we discuss are located Of course, it's so only when full-profile meniscuses are kept in pores even on the maximum moistening level, but this is within already explicited boundary conditions.

Crystallization in the surface zone of plastic substrates.

A very different situation appears when we discuss crystallization on the surface of massive slay sediments in a period of their permanent drying (photo 3). Most common aggregates are acicular crystals or twins, 0.5-15mm thick, and up to 90 cm long, extruded from the clay. Less common are antholite-like fibrous packets. Inside the clay sediment a lot more aggregates may be found, displaying a contiguous cycle of transient formations from needles to splitted needles, then to satin spar micro veins, then to linear-blocked crystals, then to euhedral crystals, then back to needles. The best description of such aggregates morphology is presented in [1], where about 40 of their morphological types are described. With this, some of the aggregates have poorly described structure, and only general genetic ideas. We'll fix our attention on two of them - extruded needles, and antholite-like packets.

Photo 3a. Gypsum needles. Splitting zone is seen on the left, two types of skeleton growth - in the middle and in the right. Left needle is 1,5 cm thick. Cupp-Coutunn cave, Dvukh Kolodtsev chamber. Photo: V.Maltsev.

a) Needles. They are really not acicular crystals, but filamentary crystals, transformed into a needle as a result of clay plasticity. The capillary pressure and the crystallization pressure together press the substrate, leaving "alive" only the pores, that are optimal in the pair pressure/speed. The other pores collapse. This is a very specific type of even not geometrical selection, but some other concurrent individuals selection mechanism, not described in literature. The solution, pressed out from the collapsed pores, appears in the active pores between the crystallization front and the surface, thus providing thickness increase of the growing filamentary crystals. This can be seen from the transient zone near the roots of the needles, displaying transformation from a pour geometry controlled filamentary crystal to a faced needle, and having 1-5cm long (photo 3). As a result of this mechanism, several interesting features appear:

- the needles never grow close to each other - distances between each pair of needles vary from 1mm to several 1-2, that is several orders greater, than in a case of solid substrate;

- the crust texture has spherical symmetry with no preferences in growth directions;

Photo 3b. Antolite-like packet 2.5 cm large. Cupp-Coutunn cave, Dvukh Kolodtsev chamber. Photo: V.Maltsev.

- the individual needles parameters are very regular - in each location lengths of needles vary only within half an order, that means almost equal growing speeds (this lengths dispersion may have its explanation exclusively in non-simultaneous growth start - new pores appear while drying);

- the crystallization environment in transient zone is disbalanced - strong oversaturations, appearing when the pressed out solution (under high pressure) reaches an active (opened) pore, have a conflict with small quantities of this pressed out solution. This disbalance leads to additional structural features of the transient zone. Closer to the root, where the solution deficit is not too strong, this provides the thickness increase mostly not through the faces development, but through the crystal splitting. At some greater distance from the root, where the solution deficit comes strong, features of skeleton growth appear, up to appearance of hollow "case crystals". These two effects can be seen from the photo 3, and are both in good accordance to the theories of crystals splitting and skeleton growth [3,11].

b) Antholite-like packets. On the first look, these aggregates are very similar to usual antholites (splitted packets of filamentary crystals), but in reality each such packet is a single splitted filamentary crystal. The genetic mechanism is almost the same, that for the needles. The difference is in higher clay plasticity, allowing pressing out all the solution from the collapsed pores very close to the crystallization front. This leaves the zone of splitted growth, but eliminates the zone of skeleton growth, so we receive not faced aggregates. The screw dislocations, typical for gypsum, are not suppressed by skeleton growth, as in the previous case, but lead to the whole aggregate curving. In spite of full visual identity to antholites (the delicate splitting zone is usually destroyed while extracting from the clay), there is a way of distinguishing them. As it's noted above, curving in these two cases come from different reasons. An antholite is curved because of mechanical reasons - the growing speeds dispersion between the pores. The curvity radius shows only some general trend, and on micro level this leads to great internal tensions, and they have their result in that the fibers of an antholite are always broken on curves. An antholite-like packet has the reason of curvity in screw dislocation, balanced between micro- and macro- levels, thus having no such breaks of fibers. This difference may be always seen under 20x-50x lens or microscope.

We'll explicit, that these two types of formations were called "aggregates" only to make it more easy to understand. In strict onthogenetic terminology, they are certainly individuals, and not aggregates - they grow from a single embryo, and the crystallization process was a single-act process, that is seen from all their features. In some cases, of course, seasonal effects are also present, the evidences are sometimes corroded roots of needles and sometimes observed zonality, but they have secondary importance. The main crystallization process is contiguous - in conditions, suitable for the filamentary crystals growth, massive clay sediments are in the state of general near-permanent drying (see above), and their capacity for sulfates and water is enough for having no solution supply.

A special case: gypsum endings.

Filamentary crystals, transformed into something, having morphology far from to be filamentary or fibrous, are not specific for plastic substrates, and may form not only individuals, but real aggregates also. Let's discuss one of such speleothems, that are known as "gypsum endings", and strucrurally may be described as quasi-epitaxial coatings.

Photo 4. Quasi-epitaxial coatings (gypsum endings). Cupp-Coutunn cave, Dikobrazii chamber. The large crystal is 4.5cm large. Photo: V.Maltsev.

These rare formations, known only from two chambers in Cupp-Coutunn cave (Dikobrazii and Nizkii), at the first look display nothing filamentary or fibrous. They have extremely simple structure - a calcite-aragonite pseudo-helictite (also known as quill anthodite) [4,9] has oriented macrocrystalline gypsum overgrowth on it's surface. The very end of the pseudohelictite is completely inside one of gypsum crystals. The gypsum crystals are aligned to have L2 perpendicular to the pseudohelictite axis (photo 4). Gypsum epitaxy upon polycrystalline calcite surfaces is a nonsense, so let's try to find some explanation of such growth through the mass-transportation symmetry features.

The most evident (and certainly having its place) mass-transportation factor is the moving and evaporating capillary film on the surface. On the macro level the texture symmetry of the coating corresponds the mass-transportation symmetry in such film - the coating comes thicker while closer to the pseudohelictite end (photo 4), thus displaying the cone symmetry [13]. On some lower examination level the picture changes. Each sharp crystal edge or top, when outstanding from the surrounding, in this case must generate an aggregate, also having the cone symmetry. Such aggregates are known as crystallictites [10], and are well-studied. Here we can't see any features, necessary for crystallictites. The growth is not dendritic, there is no specific geometrical selection, the crystals aren't enlonged, the faces aren't curved. On the individuals level the coating texture show spherical symmetry, excluding growth from the capillary film evaporation, as well as growth from gravitation-controlled solution flows.

Photo 5. Isolated crystals of re-crystallized gypsum on a pseudohelictite. The crystals are up to 1cm large. The corrosion sculpture of the pseudohelictite surface is typical for the sulfuric corrosion. Cupp-Coutunn cave, Dikobrazii chamber. Photo: V.Maltsev.

So, we must search for some local features of mass-transportation through the capillary film, providing a) oriented crystal embryos generation, and b) spherical symmetry of growth on the individuals level. There are not much possible local effects - only the chemical composition dispersion and the feeding dispersion, both with many variants of changes frequency.

To begin, let's note, that various gypsum coatings on carbonate speleothems are very typical for Cupp-Coutunn - we discuss here only their unusual shape. But that's the reason for trying to answer the two formulated question separately, searching for each of these two features in other types of gypsum coatings.

With such soluble material as gypsum, any changes of toe solution chemical composition may lead to some re-crystallization. Mostly it results in crystals quantity decrease, and survived crystals dimensions increase. The Moroshkin's idea that the capillary film environment has too low mass-capacity for any re-crystallization [10], probably is true only for carbonates. For sulfates, 0.1mm film has rather high mass-capacity, and is already gravitation independent. Any way, the gypsum crystals, shown on the photo 5 (also growing on a pseudohelictite), are a result of re-crystallization, and certainly in the capillary film environment. The evidences are: a) growth/dissolution directions, controlled only by crystallography, not by any mass-transportation symmetry; b) very rare crystal embrios; c) nothing like geometrical selection; d) no gravitation control; e) corroded pseudohelictite surface; f) no possibility of flooding.

Photo 6. Intermediate type of gypsum coating, displaying the crystals morphology as on the photo 5, and the coating morphology as on the photo 6. The crystals are up to 1.5cm large. Cupp-Coutunn cave, Dikobrazii chamber. Photo: V.Maltsev.

The quasi-epitaxial coatings have several similarities with this clear case (we even can find some intermediate formations between them, like shown on photo 6), so they also may be considered as something re-crystallized. But this something must have a property of oriented crystals embrios, making difference to cases from photoes 5,6. Moroshkin [10] writes, that the property of oriented embrios and suppressed geometrical selection may appear in crystallictites, and that he received it in one of his experiments. Sletov in [12] disproves this possibility. Probably, the Stepanov's explanation [pers.comm.], that in this experiment Moroshkin used a porous substrate, and so received not clear crystallictites, but crystallictites, growing upon filamentary efflorescences, having less embryos orientation dispersion, seems to be convincing.

The last idea is a good key for understanding our coatings. The surface layer of a pseudohelictite is an enlonged calcite corespherolite in terminology from [2]. If we imagine, that its individuals grow poorly connected, with pores left between them, these pores must be isometric in the cross-section, and strictly perpendicular to the surface.

Photo 7. Filamentary gypsum crystals, growing on pseudohelictites. Crystal lengths up to 4cm. Cupp-Coutunn cave, Dikobrazii chamber. Photo: V.Maltsev.

And if there exists a long dry period in the cave, and if the bacterial sulfur processes are active while this period, then, as it's described in the section on solid substrates, it's possible to receive filamentary crystals growth in these pores, in all the cases oriented just like the crystals from the quasi-epitaxial coatings. Of course, there are too many "if" in this conception. Studies, carried out near the areas where the quasi-epitaxial coatings are found, eliminated these ifs completely. There were found several places, where the dry period continues now, and the modern filamentary crystals growth from the pseudohelictites surfaces was observed (photo 7). Now we can build a complete genetic model for these coatings:

- firstly a pseudohelictite with abnormally poor connection between individuals of the outer layer are developed;

- on the next stage the microclimate goes much drier, the pseudohelictite itself stops developing, and filamentary gypsum crystals start growing from the pores;

- finally, the microclimate goes humid again, and the moving capillary film appears again. Filamentary gypsum crystals partly dissolve, and partly appear as embryos for isometric gypsum crystals growth. Additional quantities of gypsum (it's evident, that the filamentary crystals do not contain enough material for a coating like that on photo 4), may be brought during several processes, for example with the water film on this stage.

Some additional discussion.

The suggested model of feeding of the aggregates from the antholite crust has one interesting consequence. Probably, good estimations for growth speeds and aggregates ages may be received in a significant part of cases without isotopic studies. In most of the cases the buffering zone of the substrate is only several centimeters thick. And both it's overdrying or overmoistening will cause visible crystallization breaks. Generally, for aggregates without such breaks, year-to-year irregularities of the water level range in the substrate must be within half an order (with statistical compensation during the long period). So, the growing speed for such aggregates is close to be constant, and it can be measured from one or several years (according to the precision needed) in an indirect manner - by measuring the total evaporation/condensation balance with it's re-calculation into the growth speed through pours dimensions and the mineral solubility.

Finally, let's list the filamentary crystals and fibrous aggregates types, for which the modelling from above can't be applied:

a) growing on sulfate substrates - because of different pores structure and another chemistry of the buffering zone;

b) growing inside the substrate - because of another feeding physics;

c) non-sulfate - because of another chemistry, sometimes - another physics (for example, ice has another crystallization physics);

d) growing outside caves - because of another microclimatic conditions with mostly shorter cyclicity.



1. R.Casali, P.Forti. I cristalli di gesso del Bolognese // Speleol. Emiliana, ser.2, v.1, no.7, p.1-24.

2. A.A.Godovikov, O.I.Rypenesh, V.I.Stepanov. Spherolites, spherocrystals, spheroidolites, corespherolites (in Russian) // Novye Dannye o Mineralakh (New Data on Minerals), vol. 36, 1989, Moscow, "Nauka", p.82-89.

3. D.P.Grigorjev, A.G.Jabin. Minerals onthogeny. Individuals (in Russian). Moscow, 1975.

4. C.Hill, P.Forti. Cave minerals of the world. NSS, 1986, 238 p.

5. V.V.Korshunov, V.A.Maltsev, A.A.Semikholennykh, E.B.Teploukhova. A model of non-traditional soil generation in karst caves, with the Cupp-Coutunn cave as an example (in Russian) // In book: Sovremennye problemy pochvovedeniya i ekologii. Moscow, MGU, 1994.

6. M.N.Maleev. Properties and genesis of natural filamentary crystals (in Russian) . Moscow, "Nauka", 1971, 180p.

7. D.I.Malishevsky. The growth mechanism for minerals in case of filtration through porous environments: membrane effects controlled formations (in Russian) // Zapiski Vsesouznogo Mineralogicheskogo obschestva, Sankt-Peterburg,, currently waiiting for printing.

8. V.A.Maltsev. The influence of season changes of the cave microclimate to the gypsum genesis//Proc.10th Int.Cong.Spel.Vol. III.Budapest,1989, p.813-814

9. V.A.Maltsev, C.A.Self. Cupp-Coutunn cave system, Turkmenistan, USSR//Proceedings of Bristol University speleological society, 1992, vol.19, p.117-150.

10. V.V.Moroshkin. On genesis of crystallictite type of aggregates (in Russian). // Novye Dannye o Mineralakh SSSR (New Data on Minerals in USSR), vol. 25, 1976, Moscow, "Nauka", p.82-89.

11. G.V.Russo. Splitting of gypsum crystals (in Russian) // Zapiski Vsesouznogo Mineralogicheskogo obschestva, Leningrad, 1981, vol.110, No.2, p.167-171.

12. V.A.Sletov. On onthogeny of crystallictite and helictite aggregates of calcite and aragonite from the caves of Southern Fergana (in Russian) // Novye Dannye o Mineralakh (New Data on Minerals), vol. 32, 1985, Moscow, "Nauka", p.119-127.

13. V.I.Stepanov. Crystallisation processes periodity in karst caves (in Russian) . Trudy mineralogicheskogo muzeja imeny Fersmana. Moscow, 1971, No.20., p.161-171