m^ Bulletin ^-.k 'Ji' Series E, Chemistry and Physics, 36 l)l^:i*ARTMENT OF THE INTERIOR UNITED STATES GEOLOGICAL SURVEY CHARLES D. \VAL(;OTT, Dikectok THE ACTION OF AMMONIUM CHLORIDE UPON SILICATES w:r.j^nj^ ^^^^igj-gmjES^u^orth cl^rke GrKORCS^E STIGI&ER WASHINGTON GOVERNMENT PRINTING OFFICE •1902 u...t- SEP 3 0 CONTENTS. Page. Introductory statement 7 Analcite . 8 Lencite . . . _ . 16 The constitntion of analcite and leucite 17 Pollucite 21 Natrolite 23 Scolecite , 24 Prehnite ._ i 25 The trisilicic acids 26 Stilbite 29 Heulandite , 31 Chabazite 32 Thonisonite 34 Lanmontite 35 Pectolite 36 Wollastonite 39 Apophyllite j . . 39 Datolite . _. 40 Elseolite ■ 40 Cancrinite 41 Sodalite .. 43 The feldspars 43 Olivine 44 Ilvaite 44 Riebeckite (?) 45 ^girite 46 Calamine . . . , -^.-.^ ^ 47 Pyrophyllite •_ 49 Serpentine 51 Phlogopite 51 Lenchtenbergite 52 Xanthophyllite 53 The action of ammonium chloride on rocks 53 Summary 57 LETTER OF TRANSMITTAL. Department of the Interior, United States Geological Survey, Washington, D. C. , Septeniber 26, 1902. Sir: I have the honor to transmit herewith a memoir by Messrs. Clarke and Steiger on the action of ammonium chloride upon silicates, with the recommendation that it be published as a bulletin. These researches are of great geological importance for the light they throw upon the rational constitution of minerals. They are based on a method which is wholly novel and which is capable of wide applica- tion. The work is most creditable to the authors and to the United States Geological Survey. Very respectfully^ your obedient servant, George F. Becker, Geologist in Charge, Division of Chemical and Physical Research. Hon, Charles D. Walcott, . Director United States Geologiccd Survey. 5 Digitized by the Internet Arciiive in 2010 with funding from Boston Library Consortium IVIember Libraries http://www.archive.org/details/actionofammoniumOOclar THE ACTION OF AMMONIUM CHLORIDE UPON SILICATES. By Frank Wigglesworth Clarke and George Steiger. INTRODUCTORY STATEMENT. In a series of investigations by Clarke and Schneider, which wer^. carried out in the laboratory of the United States Geological Survey between the years 1889 and 1892," a number of reactions were studied which shed some light upon the constitution of the natural silicates. Among these reactions two were of x)eculiar interest, on account of their simplicity and the ease with which they could be applied. First, in the case of talc, it was found that one-fourth of the silica could be liberated by ignition; and that the fraction thus set free was measurable by solution in aqueous sodium carbonate. This reaction suggests that other acid metasilicates may behave in a sim- ilar way, and that we perhaps have a means of discrimination between such salts and other compounds which simulate them. In other words, an acid'metasilicate may be experimentally distinguished from a pseudo-metasilicate by the way in which it splits up when ignited. Evidence bearing upon this supposition will be found in the present paper. The second of the reactions just referred to is that between dry ammonium chloride, at its temperature of dissociation, and various silicates, different minerals being very differently attacked. Some are completely decomposed, others are affected but slightly, and in certain cases substitutions are produced of a most suggestive char- acter. To a certain extent, the two reactions overlap; that is, each one bears somewhat upon the other, and hence both have received consideration in the present series of researches. In the earlier stages of our work the several silicates which were studied were heated with dry ammonium chloride in open platinum crucibles. The temj)erature chosen was 350°, at which point the chloride breaks up into gaseous hydrochloric acid and free ammonia, "Bulls. U. S. Geol. Survey No. 78, p. 11; No. 90, p. 11; No. 113, pp. 87, 34. 8 ACTION OJF AMMONIUM CHLORIDE ON SILICATES. [bull. 207. and in this way partial changes were effected. Later, the heatings were performed in sealed combustion tubes, and then the reaction proved to be much more far-reaching. In nearly every case the material taken for investigation was ground up into one large, uni- form sample, upon which all the experiments were performed, and in that way the results obtained are comparable with one another. The few exceptions to this rule of procedure will be noticed at the proper places. In testing for soluble silica, a standard solution of sodium carbonate, containing 250 grams to the liter, was used, and here again the experimental conditions liave been kept uniform. So much premised, we may proceed to the description of our investiga- tions, species by species, in detail. ANALCITE. Analcite, from many points of view, is a species of X3eculiar inter- est, and of late j^ears it has received a great deal of attention. Its formula may be written in various ways, especially as regards the interpretation of its one molecule of water; but evidence too often has yielded before preconceived opinion. Additional evidence is now available, partly from the experiments of Friedel, and partly from the data obtained during the present investigation. The analcite first examined by us was in well-developed crystals from Wassons Bluff in Nova Scotia. A uniform sample was pre- pared, as usual, and the analysis, given below, is contrasted with the theoretical composition required by the accepted empirical formula NaAlSigOg . HgO. Found. Calculated. SiOa ... . 57.06 21.48 .13 .16 12.20 .58 8.38 54.55 A1203 23.18 Fe.,0, CaO Na^O 14.09 H^Oat 100° H^O over 100° -. .. 8.18 99.99 100.00 Fractions of ivater. At 100° 0.58 At 180° . 1.16 At 260° 3.64 At 300° 1.57 Low redness. ." 1. 90 Full redness .11 Blast none 8.96 CLARKE AN STEIGER. "] ANALCITE. 9 The fractional water determinations were made by heating in an air bath to constant weiglit at each teniperatnre np to 300°, and finally over the direct flame. The first fraction, at 100°, is evidently hygroscopic or extraneous water, which can be disregarded. The remainder of the water, 8.38 per cent, belongs to the species. The significance of the analj^tical figures will be considered later. Upon boiling the powdered analcite with the standard sodinm car- bonate solution, 0.73 per cent of silica was extracted. After ignition the mineral in two determinations yielded 1.46 and 1.38 per cent, respectively. The splitting off of silica is, therefore, very slight; and one of the formulae proposed by Doelter,'^ NagAlgSiaOg+SHaSiOg, may be set aside as improbable. Metasilicic acid, or an acid metasilicate, can hardly be present in analcite ; although the possibility of a neutral metasilicate, as indicated by the empirical formula, is not excluded. If Doelter's formula were correct, one-half of the silica should be liberated by ignition. Upon heating analcite with dry ammonium chloride, notable results were obtained even in an open platinum crucible. Sodium chloride was formed, which could be leached out by water and measured, while ammonia, free from chlorine, was retained by the residue to a notable and surprisingly stable degree. The experiments in detail were as follows: A. Analcite, mixed with four times its weiglit of ammonium chloride, was heated for four hotirs to 350°. There was a gain in weight of 2.18 per cent, and 6.10 per cent of soda, or one-half of the total amount, was converted into NaCl, which was leached out by water, examined as to its purity, and weighed. In the residue 1.20 per cent of silica was extracted by sodium carbonate, showing that no more splitting off had occurred than was previously observed. The gain in weight, as will be seen from subsequent experiments, is due to the fact that all of the NH4CI had not been driven off, or else that more water was retained. B. Analcite was ground up with four times its weight of NH4CI, heated for several hours, reground with another fourfold portion of chloride, and heated to 350° for twenty-one hours. Grain in weight, 0.08 laer cent. 5.57 per cent of soda was extracted as chloride. C Analcite heated to 350° for eight hours with four times its weight of NH^Cl. Loss of weight, 0.10 per cent. D. Six grams of mineral and 28 of chloride, mixed by thorough grinding, were heated to 350° for fourteen hours; then were reground with 28 grams of fresh NH^Cl and heated for thirty-five hours. Loss of weight, 0.18 per cent. 5.07 per cent of soda was extracted as chloride, plus 0.14 of ammonium chloride unexpelied. 2.03 per cent of silica was rendered soluble in sodium carbonate. So far three facts are noticeable. First, the weight of the mineral after treatment is almost exactly the same as before, showing that gains and losses have balanced each other. Secondly, little silica has been split off. Thirdly, approximately, but not rigorously, one-half of the soda has been converted into NaCl. In A it was exactly half; in the other experiments, a little less than half. Furthermore, in the sodium chloride dissolved out, there is only a very little ammonium aNeiies Jahrbuch, 1890, Vol. I, p. 133. 10 ACTIOIT OF AMMONIUM CHLOEIDE ON SILICATES. [bull. 207. chloride, amounting at most to 0.14 per cent, calculated upon the weight of the original mineral. In the residue of the analcite after extraction of sodium chloride, abundant ammonia can be detected, with either no chlorine or at most a doubtful trace. If, however, the unleached mineral, still retaining its sodium chloride, be heated strongly, ssby, from 400° up to redness, NH4CI is regenerated and given off. Its absence, as such, both from the leach and the residue was repeatedly proved. The ammonia and water retained by the analcite after healing to 350° with ammonium chloride were several times determined, and the following percentages, still reckoned on the original mineral, were found : NH3. H 0. InB .- 2.03 2.19 2.36 2.35 2.06 2.25 InC - 2.00 InD - --- 1.89 (( << Mean - - - - - -- 2.20 2,04 Correcting the ammonia for the 0.14 of IsrH4Cl found in D, the mean value becomes 2.15. The determinations of it were made by three distinct methods, and there is no possible doubt as to its presence. The composition of the analcite after the treatment with ammonium chloride may now be considered, with the subjoined combination of the data. The ISTaCl in A, 11.50 per cent, was in material which had gained 2.18 per cent, and is subject to a correction which reduces the figure to 11.26. In B, C, and D the corresponding correction is so small that it may be neglected. The last column gives the composi- tion of the leached residue, recalculated to 100 per cent, after deduc- tion of NaCl and the soluble silica. The letters refer back to the several experiments, and the little iron is included with the alumina. A. B. C. D. Average. Residue. Sol SiOa 1,20 2.03 54.96 21.37 .16 9.57 7.12 2.21 1.89 1.61 54.96 21.37 .16 10.44 7.12 2.15 2.04 Insol. SiO^ 62.59 AloOo 34.34 CaO .18 NaCl 11.26 10,50 Na^O ' - -- 8.11 NH3 2.03 2 25 2.19 2.00 2.46 H2O 2.32 99.31 99.85 100, 00 '\^iVr] ANALCITE. 11 The results thus obtained with aualeite from Nova Scotia were so remarkable that further investigation seemed to be needed upon material of different origin, and with variation in the details of manipulation. The new experiments, which have led to highly inter- esting consequences, are now to be described. To the kindness of President Regis Chauvenet, of the State School of Mines, we are indebted for a liberal supply of well-crystallized analcite from North Table Mountain, near Golden, Colo., of which a uniform sample of about 80 grams was prepared. An analysis of the mineral gave the following results: SiO^ 55.72 AI2O3 23.06 CaO .17 NaaO , 12.46 H^Oat 100° 0.13 H2O above 100° _ : 8. 26 99. 80 Water by fractions. At 100° 0.13 At 180° .75 At 260° :..-. 2.44 At 300° 1.28 At 350° 1 1.76 At redness . . 2. 03 8.39 Above a low red heat no f urtliei- loss of weight was observed. Upon boiling the powdered mineral for fifteen minutes with the standard solution of sodium carbonate, 0.45 per cent of silica was dissolved. After ignition, 0.57 per cent was soluble, which is practically the same amount. No silica was split off by heating. The experiments with ammonium chloride fall into two series. The first of these was conducted precisely as in the case of the Nova Scotian material, namely, hy grinding the powdered mineral into an intimate mixture with four times its weight of the chloride, and heating in an open crucible. In three cases the material, after volatilization of the ammonium chloride, was reground with a fresh amount of the salt, and then heated again. The temperature and duration of the experi- ments were purposely somewhat varied. After heating, the material was leached out with water, the sodium chloride which had been formed was estimated, and in the residue the fixed ammonia was determined. In this series there were four experiments, with results as follows : Hours heated. Temper- ature. Soda re- moved. Ammonia in residue. A 28 8i 300 350 4.75 6.36 2.04 B 2.88 C 26 350 3.76 1.72 D 5 340-380 6.70 2.85 12 ACTION OF AMMONIUM OHLOEIDE ON SILICATES. [bull. 207. In the aiialcite from Nova Scotia the ammonia retained by the leached residue ranged from 2.03 to 2.36 per cent, while the extracted soda varied from 5.07 to 6.10. In two of the new exi)eriments these figures are perceptibly exceeded, and they represent the shortest duration of heating. Prolonged heating seems to be undesirable, and seems to undo a part of the reaction which has taken place; otherwise the results obtained are of the same order as their predecessors. About one-lialf of the soda in the analcite is converted into chloride, while variable ammonia is retained. In the second series of experiments a sealed tube was substituted for the open crucible. The powdered analcite was intimately ground with four times its weight of ammonium chloride, as before, and then heated to 350° in a tube furnace for from four to eleven hours. Under these conditions practically the whole of the soda in the mineral was converted into sodium chloride, while all of the liberated ammonia was absorbed by the residual silicate. Upon leaching the contents of the tube with water, to remove sodium and ammonium chlorides, a residue was obtained which exhibited constant composition whether dried at 100° or at the ordinary temperature of the air. Three samples of the residue were prepared and analyzed; other samples were partially examined and used for subsidiary experiments. The three analyses, lettered for future reference, were as follows, the analcite itself being included in the table for comparison : Analcite. Residue A. Residue B. Residue C. SiO^ .--- AI.Oo ... ... 55.72 23.06 .17 12.46 61.93 25.21 61.68 25.88 61.79 25.24 CaO - . ..;... ]Sra.,0 .40 7.28 4.50 .22 6.95 4.91 .28 NH3 '. ... .. 7.71 H20 .. ... 8.39 5.01 99.80 99.27 99.09 100. 03 Residue C was prepared with the greatest care, and was air dried. Exposed over sulphuric acid in a vacuum desiccator for fourteen days, it lost in weight onlj^ 0.08 per cent. Tested for chlorine, only a slight trace could be recognized, but upon boiling for fifteen minutes with sodium carbonate solution it yielded 1.97 of soluble silica. After ignition only 1.70 of silica Avas soluble, or somewhat less than before. Upon heating to constant weight at 300°, only 0.46 per cent was lost, but at 350° it slowly decomposed, giving off ammonia. At 300° the compound is stable. The 0.28 per cent of soda remaining in residue C may be regarded as representing unaltered analcite, doubtless coarser particles which CLARKE and! STEIGBR. J ANALCITE. 13 escaped complete transformation. It corresponds to 1.08 per cent of analcite, which, together with the 1.97 of soluble silica and tlie 0,46 of water lost below 300°, may be deducted from the substance in order to obtain the composition of tlie definite compound. The latter amounts to 94.72 per cent of the total residue, and agrees very nearly in composition with the formula 2NH3.H20.Al203.4Si02. Recalculating the 94.72 of residue to 100 per cent, we get the follow- ing comparison between analysis and theory : SiO,- A1,0,, H,0. Found. 61.07 26. 15 8.14 4.64 100. 00 Calculated. 60.92 25. 88 8.63 4.57 100. 00 Written in rational form the compound becomes equivalent to an anhydrous ammonium analcite, NH.AlSiPe; that is, analcite in which sodium has been replaced by ammonium. From this point of view the reaction between analcite and ammonium chloride becomes a simple case of double decomposition, and is per- fectly intelligible. To establish this conclusion, however, corrobora- tive experiments were necessary. In the first place, the observed equivalency between the sodium lost and the ammonia gained might be due to a mere coincidence, and so far be illusory. One atom of sodium, taking chlorine from ammo- nium chloride, liberates one molecule of ammonia, the amount which the analcite residue has retained. Suppose more ammonia were pres- ent; could it be absorbed? To answer this question another tube was prepared, with the usual mixture of analcite and ammonium chloride. This was covered by a loose plug of glass wool, in front of which we placed enough pure lime to liberate about double the normal amount of ammonia. The tube was then sealed, and heated to 350°, as in the previous experi- ments. Upon opening the tube, a strong outrush of ammonia was noticed; but in the leached and thoroughly washed residue, only 7.52 per cent of ammonia was found. This quantity agrees with that from the previous samples, and shows that the limit of the reaction has been practically reached. One molecule of ammonia is retained, and no more. Still another experiment was tried upon a portion of the residue marked C. If the compound is really an ammonium salt, it should 14 ACTION OF AMMONIUM CHLOKIDE ON SILICATES. [bull. 207. be decomposable by caustic soda in such a way as to reverse the reac- tion by which it had been obtained. The substance, however, is very insoluble, so that the reaction takes place slowly. To phenol phthal- ein it is absolutely neutral, and with Nessler's reagent it reacts only after long standing. To settle the question a weighed portion of the compound was boiled in a distilling flask with a 10 per cent solution of sodium hydroxide, to which water was added from time to time. The distillate was col- lected in a tube containing aqueous hydrochloric acid; and the ammonia which passed over was weighed, ultimately as chloroplati- nate. Bj^ four hours' boiling 6.90 per cent of ammonium was driven off and determined; and the residue remaining in the flask, after washing until no alkaline reaction could be detected in the wash- water, was examined for soda, of which 10.41 per cent was found. The anticipated reaction had taken place, although not completely; it was enough, however, to confirm our opinion, and to establish the nature of the new compound beyond reasonable doubt. Other con- firmation was obtained later, from the study of leucite. The foregoing paragraphs now enable us to understand a phenome- non which we observed in our work with the open crucible. In that case a partial reaction takes place between the analcite and the ammonium chloride, producing, as in the sealed tube, a mixture of an ammonium alumino-silicate with sodium chloride: the two sub- stances being separable by leaching. But if, instead of leaching, the mixture be heated to full redness, ammonium chloride is re-formed and given off, leaving a residue which contains little or no sodium chloride, and is wholly insoluble, or almost so, in water. That is, the reaction which occurs at 350" is reversed at the higher temerature, and anhydrous analcite, or an isomer of it, is regenerated. Ammo- nium and sodium again change places, and the original state of molec- ular equilibrium is restored. What, now, is the nature of the product obtained in the open cru- cible after sodium chloride has been removed? Is it a definite inter- mediate compound or an indeterminate mixture? At first we were inclined to accept the first of these alternatives, and we assigned to the substance the formula IIaNa3Al4Sig024.NH3, in which the ammonia plays a part equivalent to that of water. In this expression we were influenced by the researches of Friedel," who had shown that ammonia could in part replace the "zeolitic" water of analcite; but it now appears that the phenomenon observed by him is quite distinct from that discovered by us, and is, indeed, of an entirely different order. We may, therefore, in accordance with our new data, rearrange the formula, transforming it to that of an ammonium salt, HNa2]S'H4Al4 Si8024, the agreement with the analytical figures being approximate only. The results obtained are not sharp enough for certainty. This product we are now inclined to regard as a mixture, although a Bull. Soc. min. Prance, Vol. XIX, p. 94, 1896, CLARKE AND"! STEIGER. J ANALCITE. 15 it is not strictly intermediate between analcite and its final ammoniiini derivative. Only half of the eliminated sodium has been replaced by ammonium, while hydrogen, or water, makes up the deficiency. It seems probable that the reaction in the sealed tube and that in the open crucible are at first essentially the same, but that in the latter case secondary reactions follow, which cause the variations in the final riesults. In the sealed tube the element of pressure comes into play, and the reaction is complete. In the open crucible pressure is lacking; some ammonia escapes fixation and reacts upon a jjart of the sodium chloride which was at first formed ; hence the composition of the leached residue is essentially modified. This residue may be a definite compound, but the case in its favor is unproved and the presumption is rather against it. The most remarkable fact developed by the foregoing experiments is the easy replaceability of the soda in analcite. This replaceability, however, is not limited to the substitution of ammonium for sodium; it appears to extend to other bases as well, and this we have proved in the case of silver. This is illustrated by three experiments upon the Colorado analcite, as follows: A. Analcite, intimately mixed with dry silver nitrate, was heated in a sealed tube to 400° for four hours. B. Analcite and silver nitrate were heated in a sealed tube to 250° for four hours. C. Ammonium analcite, mixed with dry silver nitrate, was heated in a sealed tube to 250° for four hours. All the products of these heatings were leached with water, and washed until the filtrates gave no test for silver; the residues were then dried on the water bath. The product in each case was a white powder not differing in appearance from the original mateKal. The analyses of the different portions are given below, together with the composition of the theoretical compound, AgAlSigOg.HaO, which is given in column D. A. B. C. D. SiOa 41.31 16.44 37.45 .85 4.29 40.08 16.29 36.91 .ai 5.86 42.69 18.22 32.01 .68 6.08 .69 none 39.35 A1203 - - 16.72 AgjO . 38.03 NazO H,0 - .. 5.90 NH3 Nitrates _ . ._ none none 100. 34 99.95 100. 37 100. 00 From preparation A, 13.13 per cent of the soda in the original min- eral was found in the leach water; and in B, 12.57 per cent. These quantities are slightly in excess of the amount actually present in the analcite, for the reason that a little other material which passed into 16 ACTION OF AMMONIUM CHLORIDE ON 8ILICATES. [bull. 207. the filtrates was not separated from the soda. It is enough to show that a true silver analcite has been formed, and that the transforma- tion is practically complete. A similar reaction takes place between silver nitrate and chabazite, but the product as yet has not been exhaustivel}^ examined. The reaction, it will be observed, is analo- gous to that by which silver ultramarine is produced, and it suggests a promising line of experimentation for the future. LEUCITE. Between analcite and leucite the closest analogies have long been recognized. The two minerals have similar composition, they resemble each other in crystalline form, and they yield, upon alteration, prod- ucts of the same order. Recently also, analcite, like leucite, has been identified as a not uncommon constituent of volcanic rocks; analcite basalt being a good example. In view of these resemblances it was plainly desirable to compare the minerals by means of the ammonium chloride reaction, a task which has been performed with satisfactory results. In a preliminary experiment a sample of leucite taken without regard to purity was heated with ammonium chloride to 350° in a sealed tube. Potassium chloride was formed corresponding to 18.06 per cent of potash, and in the leached residue 6.90 jjer cent of ammonia was found. The foreseen reaction had occurred, and more careful work was accordingly undertaken. Our material consisted of a large, irregular crystal of leucite from Vesuvius, which yielded about 20 grams of the pure mineral. This was ground to a uniform samj)le, and a portion of it was analyzed; the analysis will be given presently. The sealed-tube experiments were conducted precisely as in the case of analcite, and they confirmed both the preliminary test and our anticipations. Chlorides were formed equivalent to 18.53 per cent of potash, 1.08 of soda, and 0.08 of alumina; the reaction, therefore, was very nearly complete. The leached residue was then analyzed, and the data, compared with the analysis of the original mineral, were as follows: Leucite. SiO, ' 55.40 AlA ' 33.69 CaO. j .16 K.,0 19.54 Residue. 60.63 26.44 trace .50 Na/) 1.25 .25 NH;, 7.35 H.,0 .24 5.17 100. 28 100. 34 CLAIIKE AN STEIGEU "1 LEUOITE. 17 Leucite, then, gives the s;uue reaction as anal(!ite and yields tlio same ammoninm componnd. A closer agreement in the comi)osition of the latter could not reasonably be demanded. Ammonium leucite is formed in both cases by ordinary double decomposition in a state of approximate purity; the first silicate of ammonium, we think, which has ever been prepared. As a further check upon the results so far obtained, an attempt was made to transform ammonium leucite into the corresponding lime salt, CaAljSijOia, by fusion with calcium chloride. The ammonium leucite was mixed with a saturated solution of calcium chloride, which Avas evaporated to dryness, then heated gradually to dehydration, and finally fused. Ammonium chloride was given off and identified. Upon treating the fused mass with water, filtering and thoroughly washing the residue, a white powder was obtained which, after drying at 100°, was analyzed. It was also examined microscopically by Mr. J. S. Diller, who found it to consist of apparently isotropic grains, showing traces of incipient crystallization. The following analysis is contrasted with the theoretical composition of calcium leucite, from which it varies considerably. Found. Calculated. SiO, 54.35 26. 23 17.38 .16 .25 .28 1.24 60.30 Aip.. ... . 25. 63 CaO 14.07 KjO Na.,0 . CI . Loss on ignition 99.89 100.00 Evidently the desired salt was not definitely obtained, and the product appears to be a mixture. The reaction, however, tends in the right direction, and deserves further study under other condi- tions. Probably the water which was present in the mixture of sili- cate and chloride took part in the changes produced, although of this we can not be certain. It is interesting to note that the product obtained approximates in composition to the meteoric mineral mas- kelynite, which is regarded by Groth as probably equivalent to a calcium leucite. THE CONSTITUTION OF ANALCITE AND LEUCITE. In all of the earlier attempts to discuss the constitution of analcite the molecule of water which it contains has been a chief element of uncertainty. Should it be regarded as, representing hydroxyl or as 9506— No. 207—02 2 18 ACTION OF AMMONIUM CHLORIDE ON SILICATES. [bull. 207. water of crystallization? That question arose first of all. Under the first interpretation analcite became a diorthosilicate : AlNaHjSigO^; under the latter its equivalency with leucite appeared . The researches of Friedel, however, have settled this question in part, and whatever the function of the water may be it is something outside of the true chemical molecule; for all the water can be expelled from analcite bjT^ heat, without destruction of the crystalline nucleus, the anhydrous salt, and it is taken up again upon exposure of the dehydrated min- eral to moist air. But whatever its mode of union may ultimately prove to be, the amount of water in analcite corresponds to the simple molecular ratio which is shown in the ordinary formula of the species. One molecule of analcite holds a certain definite number of water molecules, and Friedel's observations are not incompatible with the idea that these are retained with varying degrees of tenacity. This idea is suggested by the various series of fractionation experiments which have been made from time to time by independent workers, even though the data are not by any means concordant. Thus Lepierre " found that half the water of analcite was driven off at or below 300°, the other half above 440°. In our own experiments three- fourths were expelled at 300°, the remaining fourth being held up to a much higher but undetermined temperature. In both series the water fractions are represented by fourths, but Friedel's experi- ments* indicate a continuity of loss in weight of a quite dissimilar order. Friedel holds that all of the water fractionations heretofore made upon analcite are fallacious, and that no definite fractions can be identified — a conclusion strongly supported by his own data, even though the proof is not absolutely positive. The most that can be said is that the weight of evidence so far is in favor of Friedel's contention, but that additional investigation is necessary in order to reconcile all discrepancies. The full significance of the water in analcite remains unknown. Eliminating the water from analcite, the empirical formulse for both analcite and leucite appear at once to be identical in form and to represent salts of ordinary metasilicic acid. Indeed, both minerals have been commonly regarded as metasilicates; but upon this iDoint the j)roduction of the ammonium derivatives now sheds a new light. In the formation of the latter compounds the fixed bases of the original salts have been replaced by a volatile base, and the substances so formed split up upon ignition in such a way as to give evidence regarding their constitution. For example, if ammonium leucite is a true metasilicate, a salt of the acid HjSiOg, it should break up, when ignited, in accordance with the following equation : 2NH4Al(Si03)2=Al2(Si03)3 + 2NH3-j-Il20-f-Si02; " Bull. Soc. chim. France, 3d series. Vol. XV, p. 561, 1896. &Bull. Soc. min. France, Vol. XIX, p. 363, 1896. CLABKE AND STEIGEB ^_^~\ CONSTITUTION OF ANALCITE AND LEUCITE. 19 that is, one-fourth of the silica, ought to be set free, measurable by extraction with sodium carbonate solution. No such splitting off occurs, however. An ammonium leueite which already contained 1.97 per cent of soluble silica gave only 1.70 per cent after ignition ; hence no additional silica had been liberated. We may conclude, therefore, that analcite and leueite are not true metasilicates, but pseudo-com- pounds, either salts of a polymer of metasilicic acid or mixtures of ortho- and trisilicates analogous to those which we find among the plagioelase feldspars and in the mica group. In order to discuss the constitution of analcite, let us recur to our analysis of the variety from Nova Scotia. It is at once evident from the comparison made on a preceding page that our sample of the mineral varies notably in composition from the requirements of theory. The silica is 2^ per cent too high, while alumina and soda are correspondingly low. No probable impuritj^ and no presumable errors of manipulation can account for so great a divergence. If we consult other analyses, as we find them tabulated in manuals like those of Dana and Hintze, we shall find other cases resembling this, and also examples of variation in the opposite direction, with silica low and an apparent excess of bases. Most analcite gives quite sharplj' the metasilicate ratios required by the accepted formula; but the variations from it are large enough, common enough, and regular enough to command attention. The analyses are not all covered bj^ the recognized theory, and the apparent irregularities are not fortuitous, but are systematic in character. One explanation of the seeming anomalies is simple and clear. If analcite, instead of being a metasilicate, is really a mixture of ortho and trisilicate, then all of the analyses become intelligible. In most cases the two salts are commingled in the normal ratio of 1:1, but in our analcite the trisilicate predominates, while in some other samples the ortho-salt is in excess. All reduce alike to the simple expression NaAlX.HgO, in which X represents nSi04+mSi308, a formula which agrees with evidence from various other sources. For example, analcite may be derived in nature either from albite, AlNaSigOg, or nephelite, AlNaSiO^, and on the other hand alterations of it into feldspars have been observed. Its closest analogue, leueite, has yielded pseudomorphs of orthoclase and elseolite, while leueite and analcite are mutually convertible each into the other. The evi- dence of this character — the evidence of relationship between analcite and other species — is varied and abundant, and the simplest conclusion to be drawn from it is that which ha;? been given. Every alteration, every derivation, every variation in the composition of analcite points to the same belief. The consistency of the data can not well be denied. 20 ACTION OF AMMONIUM CHLORIDE ON SILICATES. [bull. 207. In tlie case of a normal analcite — that is, one which conforms to the usual empirical formula — the expression which best represents these relations is Al.Na, (SiO,), (SiaOg)^. 4H2O; and leucite is the corresponding potassium salt, but anhydrous. Structurally this is comj)arable with the formulae of garnet, zunjdte, sodalite, and noselite, all of which are isometric in crystallization. The more important of the symbols are as follows : Al- / Si04 = Ca /Ca -SiO, = Ca Al- Si04=Na2 \ai-ci .SiCX=Na, Al- Si04=Na2 NA1-S04-Na -SiO,=Na, Al- Si04=Al Garnet. Si04=K.2 \Al-si04: -Si.,(L=Ko Si04=Al Sodalite. :A1 Al- \ Si04=Al Noselite. Si04=Na2 / NAl-Si04=Al — Si30s=Na2 +4H2O SigOg^Al . Leucite. SigOg^Al Analcite. That is, analcite and leucite become members of the garnet-sodalite group of minerals, and their relations to nephelite, albite, etc., natural and artificial, are perfectly clear. In analcite there may be admix- tures of strictly analogous ortho- or trisilicate molecules; but these remain to be sex3arately discovered. The ammonium salt correspond- ing to such a mixture, when ignited, might be expected to give the following reaction : Si04— Anig ^Al-SigOgEEEEAl SiO. \ Al-SigOg^Al; Al Si()4 = Ain2 -2Am20=Al- -SiO, Si.,0«-^A1 Si,0«=Al a reaction which is in harmony with our experimental results. In it no free silica appears; and manj'^, if not all, conditions of the problem are satisfied. One difficulty, however, stands in the way of an unqual- ified acceptance of these formulae. Garnet, sodalite, nephelite, albite, etc., are but moderately attacked by ammonium chloride, and so far have yielded no definite ammonium derivatives. Whether this dif- ference in behavior is constitutional or not it is hardly possible to say, but it must be taken into account in connection with all of the other evidence. We must remember, moreover, that the formulae "YS/r""] pollucite. 21 are not ultimate verities to be blindly accepted. They are simply expressions which represent composition and a wide i-ange of estab- lished relationships, and which serve a distinct purpose in tlie coi-rela- tion of our knowledge. Properlj^ used, with due recognition of their limitations, they are helpful, and suggest possibilities of research ; misused, they may become mischievous. They now satisfy most of the known conditions, and that is a sufficient warrant foi' their existence. POLLUCITE. On account of the general analogy between pollucite, analcite, and leucite, the first-named species of the three seemed to deserve some attention. Through the kindness of Prof. S. L. Penfield, about 10 grams of very pure material from Hebron, Me., was put at our dis- posal, and three analyses of it by Wells were already on record. ^^ The average of these analyses is as follows : SiOa ..... 43.53 AI2O3 , 16.3'7 CaO .22 Na^O 1.81 K2O . .49 Li^O __.- .04 CS2O 36.08 HjO... 1.52 100. 06 Five grams of the finely powdered mineral was heated in a sealed tube with four times its weight of ammonium chloride to 350° during forty hours. Upon leaching with water 0.14 per cent of CaO, 1.28 of NajO, and 12.30 of CsgO were extracted. Probably the calcium chloride formed contained some potassium chloride, but that point was ignored as irrelevant. The air-dried -residue had the following composition : SiOa . --- 49.21 AI2O3 18.32 CaO . . none CsjOCKaO) 28.84 NajO none NH3 2.52 H2O 1.91 100. 80 The high summation here is due to reckoning some KCl as CsCl. Of the silica in this product 2.36 per cent was soluble in the standard solution of sodium carbonate. After ignition, 4.13 ]3er cent was soluble. Some silica, therefore, was split off bv heating. a Am. Joiir. Sci., 3d series, Vol. XLI, p. 313, 1891. 22 ACTION OF AMMONIUM CHLORIDE ON SILICATES. [bull. 207. In a second experiment one gram of pollucite was heated with ammonium chloride for five hours, the other conditions being the same as before. Upon leaching, 11.55 per cent of CsgO was extracted, and a partial analysis of the air-dried residue gave the following data: SiO^ 47.87 AlA 17.85 NHg 2.83 H2O 1.55 Alkalies (by difference) 29.90 100. 00 The two products were evidently the same, and onlj^ about one- third of the alkalies in the pollucite had been extracted. So, also, the ammonia taken up was only about one-third of that which was retained by analcite and leueite. The transformation, then, is merely partial, and further experimentation seems to be unnecessarj^, at least for j)resent purposes. The analogy with analcite and leueite is far from perfect. NATROLITE. In a preliminary experiment upon an impure, yellowish natrolite from Aussig in Bohemia, we found that this species was peculiarly well suited to reaction with ammonium chloride. By heating with the reagent in a sealed tube and subsequent leaching with water, 17.56 per cent of bases was extracted, and in the residue 8.29 per cent of ammonia was found. Careful work upon this species was therefore desirable. The material available for our experiments came from the well- known locality at Bergen Hill, N. J., and consisted of a mass of slender needles densely matted together. Part of the uniform, ground sample was analyzed, with fractional determinations of the water, and part was used for the sealed tube experiments, precisely as in the research upon analcite and leueite. Three of these experi- ments were made, and in each case the natrolite was mixed by grind- ing in an agate mortar with four times its weight of dry ammonium chloride, after which it was heated to 350° in the sealed tube. Even during the grinding a slight reaction took place, and a distinct smell of ammonia was given off by the mixture. With pectolite the same smell was perceived. The three experiments may be summarized as follows : A. Heated eleven hours. Upon leaching, 14.89 per cent of soda and 1.20 of lime were extracted. In the residue 9.26 per cent of ammonia was found. B. Heated nine hours. Leach not examined. 9.26 of ammonia in residue. The complete analysis of the residue is given farther on. C. Heated three hours. 14.09 per cent of soda and 0.20 of lime were extracted. The residue contained 8.87 per cent of ammonia. In this instance the heating CLARKE AND "I STEIQBR. J NATJROLITE. 23 was relatively brief, in order to learn whether its duration could he advanta- geously lessened. The reaction was evidently less complete than in experiments A and B. In the subjoined table we give first the analysis of th(^ natrolite itself, and then that of tlie leaclied residue from experiment B. In tlie latter we found that 0.86 per cent of silica Avas soluble in sodium carbonate solution, and that soda and lime remained corresponding to 4.01 per cent of the original mineral. Deducting these impurities, together with the 0.42 per cent of hygroscopic water, and recalculating to 100 per cent, we get the reduced composition of the residue. In the last column is given the calculated composition of an anhydrous ammonium- natrolite, (NH4)2Al2Si30io. This compound has evidently been formed to an extent represented by over 94 per cent of the leached natrolite residue. The agreement between theor}^ and even the unreduced analysis is practically conclusive on this point. Natrolite found. Residue found. Residue reduced. (NH4)2Al2 SiaOio cal- culated. SiO. AIA .._ 46.62 26.04 1.48 none 15.67 53. 71 29.94 .34 53.86 30.52 54.06 30 43 CaO :_ _ K,0 , NajO .37 9.26 .42 5.94 NHs 9.85 10.14 HjOatlOO' .39 10.18 B..fi above 100° 5.77 5.37 100. 38 99.98 100. 00 100. 00 The fractional water determinations will be given later, in connec- tion with similar data for scolecite (p. 25). It ma,y not be superfluous to note that the Vt^ater given in the last two columns of the foregoing table represents the difference between ammonia and the hypothetical ammonium oxide which has replaced soda. Two other experiments upon natrolite remain to be noticed. First, the fresh mineral was boiled for fifteen minutes with a 25 per cent sodium carbonate solution; 0.72 per cent of silica dissolved. Similar treatment of ignited natrolite took out 0.62 per cent. No silica is split off by ignition. Ammonium natrolite before ignition yielded 0.85 per cent of soluble silica, and after ignition 0.86 per cent. Here again no silica had been split off from the molecule, and practicallj^ none was liberated by the action of the ammonium chloride upon the natrolite. A simple, direct substitution of ammonium for sodium had occurred. 24 ACTION OF AMMONIUM CHLORIDE ON SILICATES. Lbull- 207. Heated with ammonium chloride in an open crucible, natrolile gives only a partial reaction. This is shown by the earlier experiments of Schneider and Clarke upon natrolite from Magnet Cove, Arkansas, from which, by a triple heating with the reagent, onlj^ 0.50 jier cent of soda was extracted out of a total of 15.40. SCOLECITE. On account of the well-recognized analogy between natrolite and scolecite, the latter mineral seemed to be peculiarly worthy of exami- nation. The specimen at our disposal was a mass of stout, radiating needles, which was collected by one of us at Whale Cove, on the island of Grand Manan, New Brunswick. Scolecite, we believe, has not hitherto been recorded from this locality, and on this account alone the material deserved attention. Three sealed tube experiments were carried out, essentiallj^ as in the case of natrolite, as follows : A. Heated ten hours at 350°. 13.74 per cent of lime and 0.35 of soda were taken out. The residue contained 8.78 per cent of ammonia. B. Heated ten hours at 370°. 13.97 of lime and 0.22 of soda were extracted. 8.48 per cent of ammonia in the residue. On account of the excessive temperature of this experiment, some reversion of the converted material had taken place. C. Heated five hours at 340°-350°. Leach not studied. 8.91 per cent of ammonia in residue. Analyses of the scolecite and of residues B and C are given below. The less perfect transformation in the case of B is evident. Scolecite. Residue B. Residue C. SiO., . Al,Oo 45.86 25.78 13.92 .41 53.39 30. 51 .63 undet. 8.48 .74 6.38 53.69 30.50 CaO Na^O NH, _. . .43 .29 8.91 H^OatlOO". - _ _ _ .40 13. 65 .12 H^O above 100° 6.52 100.02 100. 03 100.45 The product of the reaction is plainly the same as that obtained from natrolite, and the identity in type of the two species is perfectly clear. This fact is further emphasized by an experiment upon the solubility of silica. The fresh scolecite gave up 0.36 per cent of silica to sodium carbonate solution, and the ignited mineral yielded only 0.50 per cent. Again, natrolite and scolecite behave in the same way. Ui^on both minerals fractional determinations of the water were made, and the amount lost at each temperature was noted. The CLARK STEIGER CE AND"! IGER. J SOOLECITE. 25 results, expressed in percentages of tli(^ original minerals, were as follows : Temperature. Water lost. Natrolite. Scolecite. 100°- --- 0.39 .40 • 37 8.51 .72 .12 .06 10. 57 0.40 180° - .52 250° - - . _ ... 4.76 350° - . 55 Incipient redness Full redness Over blast - - - - - - ^ _ _ ^ _ 7.72 .04 .06 14.05 Scolecite contains one more molecule of water than natrolite, and that amount, one-third of its total, seems to go off at a lower temper- ature than the other two molecules. Otherwise the two series of ex- periments are probably not far apart, and they indicate that the water is in neither case constitutional. The same conclusion is suggested by the existence of the anhydrous ammonium compound, the three formulae being as follows: Scolecite CaAl^SisOio. SH^O Natrolite Na^Al.SigOio. 2H,0 Ammoninm natrolite (NH4)2Al2Si30io The parallelism is complete ; and all three compounds are evidently salts of an acid, HgSigOio, which is probably orthotrisilicic acid, Si302(OH)8. The relations of this acid to its anhydrides will be con- sidered later. PREHNITE. In a former bulletin upon the constitution of the silicates, " one of us attempted to show that natrolite, scolecite, and prehnite were similar in chemical structure, provided that all or part of their water was regarded as constitutional. The formulae then assigned were as follows : Scolecite : - Al2(SiOj3CaH,. H.O Natrolite - AI, ( SiQi ) sNa^H^ Prehnite Al2(SiOj3Ca2H2 Two of these formulae must now be abandoned, because of the exper- imental evidence which we have obtained, but the prehnite remains to be considered. fl Clarke, F. W., Bull. U. S. Geol. Survey No. 125, p. 45,1895. 26 ACTION OF AMMONIUM CHLOKIDE ON SILICATES. [bull. 207. The material chosen for examination was an old specimen of preh- nite from Paterson, N. J. The analysis of it, with fractional water determinations, is given below : SiOj 42.31 AI2O3 - 19.95 Fejps 6.20 PeO none CaO 36.68 H^O 5.02 100. 11 Fractional water. At 100° 0.21 At 180° , - .18 At 250° .10 At 350° .11 Incipient red lieat .28 Full red heat 4.05 Over blast ^ .09 5.02 With sodium carbonate snlution, 0.38 per cent of silica was ex- tracted from the fresh mineral. From the ignited prehnite, 1.22 per cent was taken out. Very little silica, therefore, is liberated by ignition. Two determinations were made of the action of ammonium chloride, as follows : A. Heated eight hours. On leaching with water, 1.31 per cent of lime and 0.17 of alumina dissolved. B. Heated twelve hours. 1.41 per cent of lime was extracted, and in the washed residue 0.22 per cent of ammonia was foiind. Prehnite, therefore, differs widely from natrolite and scolecite in its behavior with ammonium chloride. Very little action takes place, even upon long heating to 350° in a sealed tube, and practically no ammonia is absorbed. The water is more firmly held than was the case with the other two minerals, and is almost certainly to be regarded as constitutional. The orthosilicate formula for prehnite is unaf- fected by these results, and may stand as fairly probable. Prehnite can not be correlated with natrolite and scolecite on any basis of similar chemical structure. THE TRISILICIC ACIDS. We have already shown that natrolite and scolecite are probably salts of an orthotrisilicic acid, HgSgOio, '^^ ^^id which is not particu- larly well known. As it has interesting relations to other compounds, some discussion of its constitution and its derivatives may not be out of place here. The general theory of the silicic acids is extremely simple. Silicon being a quadrivalent element, its normal acid, the orthosilicic, is '^^STEWER^^] THE TRISILICIO ACIDS. 27 81(011)4. From this, by successively eliiuinatinjj; two inoleciiles of water, two anhydrides may be derived, thus: Orthosilicic acid Si(0H)4 First anhydride, metasilicic acid 0=Si^^(OH)^ Second anhydride, silicon dioxide - . 0=Si=^0 These acids, containing one atom of silicon each, may be called the monoslllclc acids, and some of their salts are perfectly well known. Olivine and anorthlte, for Instance, are orthosllicates, while the true metasillcates are represented by talc and pectollte. The evidence in the case of the last-named mineral will be presented later. When two molecules of orthosilicic acid coalesce, with elimination of water, an orthodlsUlcic acid is formed, and this is the first member of another series, as follows: Si=(OH)s Si=(0H)3 0=Si-OH o 00 Si=(0H)3 0=Si-OH 0=Si-OH Orthodisilicic acid. Metadisilicic acid. Pyrosilicic acid. To the first and third of these acids various minerals correspond. The second acid, however, is a polymer of metasilicic acid, but differs from the latter In its possible derivatives. When an acid metasillcate is heated silica is set free, but in the case of a metadislllcate this would not necessarily occur. Possibly leucite and analcite maj^ be metadlsillcates, although the evidence so far presented does not sup- port this view. The possibility, however, we are compelled to recog- nize as one which might ultimately be verified. With the coalescence of three orthosilicic molecules a series of trisi- liclc acids begins, and one of these forms salts — the feldspars — which are the most abundant compounds existing in the mineral kingdom. The acids of the series are these : Si=(0H)3 Si=(0H)8 Si=(OH)s 0=Si-OH o 000 Si=(0H)2 Si=0 Si=0 Si=0 ' O .0 O O Si=(OH)3 Si=(0H)3 0=Si-OH 0=Si-OH Orthotrisilicic acid. Metatrisilicic acid. Trisilicic acid. Third anhydride. 28 ACTION OF AMMONIUM CHLORIDE ON SILICATES. [bull. 207. The third anhydride represents an acid to which no known salts correspond. One step further and we have a fourth anhydride, SigOg, or empirically SiOg, which may or may not be the true formula of quartz. Quartz is undoubtedly a polymer of SiOj; its most frequent associates are trisilicates — the feldspars — and hence the formula SigOg has a certain degree of plausibility. This suggestion, however, is purelj^ si3eculative and has no definite scientific value. Its validity would be most difficult to establish. From the first of these trisilicic acids natrolite and scolecite appear to be derived. If we ignore the "zeolitic water," which is not a part of the essential silicate molecule, the two compounds may be formu- lated thus: Si = 03 = Al Si = 03 = Al o o Si = 02 = Na2 Si = 02 = Ca o o Si^Og^Al Si = 03 = Al Natrolite. Scolecite. So far, no other salts of this acid have been clearly identified. The second acid of the series, like the second of the disilicic acids, is a polymer of the ordinary metasilicic compound. It is well under- stood that many so-called metasilicates are not representatives of the simple acid HgSi O3 ; some of them are mixtures of orthosilicates with salts of the third acid in this group, H4 SigOg; others may be derived from j)olymers like that which is now under consideration. For example, anhydrous analcite and jadeite are both represented by the empirical formula NaAlSigOg, but they differ widely in densitj^ in solubility, and doubtless also in crystalline form. One molecule, then, is much more condensed than the other. If analcite should prove to l)e a metadisilicate, then jadeite may be its equivalent in the trisilicic series, or it may belong with some still higher polj^mer. The possibilities are many, but to establish au}^ one of them by proof would demand more evidence than is yet in our possession. The third member of the trisilicic series is the most important of all, for among its salts are the two feldspars, albite and orthoclase, which together make up fully one-half of the solid crust of the earth. It is also noteworthy from the fact that its formula can be so written CLARKE AND STElGEll ^°] STILBITE. 29 Sis 1 = (OH), O Si= =o 0 0 = = Si- -OH as to represent two isoinei'ic forms, to wliicli distinct salts prol)Hl)ly correspond. The two formulsB are as follows: 0=Si — OH O and Si = -(0H)2 O 0 = Si — OH; and their significance is clear when we remember that the ordinary trisilicates are commonly dimorphous. Thus we have orthoclase and soda orthoclase, monoclinic; and albite and microcline triclinic; one pair perhaps belonging to one isomer, the other to the other. The rare minerals eudidymite and epididymite, which are also isomeric trisilicates, further illustrate the same conception; but we can not as yet assign either compound distinctly to either formula. By an extension of the process herein developed, which is by no means new, higher polj^silicic series may be formulated. Since, how- ever, such acids correspond to no definitely known salts, to write their formulae would be a useless exercise of the imagination. Beyond the trisilicic acids we enter the region of the unknown. STILBITE. The specimen selected for study was a nearly white, typical exam- ple from Wassons Bluff, Nova Scotia. The analysis and the fractional water determinations were as follows: Si02_. - 55.41 AI2O3 16.85 FeA .18 MgO .05 CaO ■ 7.78 Na^O 1.33 H2O 19.01 100.51 Fractional ivater. At 100° . 3.60 ■ At 180° 6.46 At 250° 3.80 At 350° 3.10 Low redness 3. 95 Fnll redness .06 Over blast .04 19.01 30 ACTION OF AMMONIUM CHLORIDE ON SILICATES. [bull. 207. On boiling with sodium carbonate, 1.37 per cent of silica went into solution. After ignition, only 1.03 per cent was obtained. No silica, therefore, is split off when stilbite is ignited. If the mineral were a hydrous acid metasilicate, H4CaAl2Si60i8.4H20, as has been assumed by some authorities, one-third of the silica should have been set free. Hence the metasilicate formula is to be regarded as unsatisfactory. The evidence here presented counts for something against it. Two samples of the ammonium chloride derivative were prepared. In leaching with water the insoluble residue was washed until the wash- ings gave no reaction for chlorine. The chlorine shown in the sub- joined analyses is, therefore, present in an insoluble form and not as adhering ammonium chloride. Dried at 50° the two products gave the following comi30sition : A. B. SiOj '._.:.... .. 60.80 18.36 1.86 .08 5.12 12. 96 1.31 60.67 ALOo -- 18.25 CaO - 1.46 Na,0 .15 NHg 5.13 H20 13.91 CI : '. 1.04 Less 0 100. 49 .29 100.61 .23 100. 20 100.38 Sample B was further examined as to the presence of soluble silica, and 1.52 per cent was found. After ignition, only 1.62 per cent went into solution. These results conform to those obtained with the orig- inal stilbite, and tend to show that the ammonium derivative is a com- pound of the same order. In the case of the unignited substance the residue remaining after the removal of soluble silica was thoroughly washed, and then examined for alkali. It was found to contain 9.30 per cent of soda, which shows that the ammonium salt had been trans- formed back into the corresponding sodium compound. From the foregoing facts it is clear that stilbite, like the zeolites previously studied, is converted by the action of ammonium chloride into an ammonium salt. That is, sodium and calcium are removed as chlorides, ammonium taking their place to form ammonium stilbite. The reaction, however, is less complete than it was in the cases of analcite and natrolite, but whether this is due to a greater stability of the stilbite molecule or only to a different degree of fineness in the powder upon which the operations were performed, we can not say. Neither have we any explanation to offer of the retention of chlorine CLARKE AND STEIGKR ] STILBITE AND HEULANDITE. 31 by the ammonium derivative. Althougli the amount of chlorine is small, it needs to be accounted for. If we discuss the comj)osition of the stilbite and of its ammonium derivative, the relations between them become very clear. Nej^lect- ing the water as "zeolitic," to use Friedel's phrase, and, therefore, as not a part of the chemical molecule, and also rejecting the 1.37 per cent of soluble silica as iprobably an impurity, the ratios derived from the anal^^sis give this empirical formula for the mineral: Na4oCai4oAl332Si9ot0246„. This corresponds to a mixture of ortho- and trisilicates in which SigOgiSiO^:: 286: 43; and uniting these radicles under the indiscrimi- nate symbol X, we have, as a more general expression, NajoCai.jo AlggaXggy ; or combining monoxide bases, which is essentially R"Al2X2. Since the Si04 groups are practically equal in number to the sodium atoms, the stilbite is probably a mix- ture, very nearly, of ]SraAlSi04 and CaAl2(Si30g)2 in the ratio of 1 : 7 This is in accordance witli the well-known theory of Fresenius as to the constitution of the iDhillipsite group, to which stilbite belongs. Stilbite is mainly a hj^drous calcium albite, commingled with varying amounts of corresponding orthosilicates of soda and lime. For the ammonium derivative similar relations hold. Taking analy- sis "B" for discussion, rejecting soluble silica and chlorine as impu- rities, and neglecting all water except that which belongs to the suj)- posable ammonium oxide, the ratios give this formula: (NH4)3oiNa4Ca2gAl358Si9>^502684. Uniting sodium and calcium with ammonium, this becomes R''357^^358(^i3<^8)314(Si04)43 ; or, more generally, ^ 357-^1358-^3575 =1:1:1- The derivative, therefore, is a compound of the same order as the original stilbite, with the ratio of 1: 7 still holding between the ortho and trisilicate groups. This conclusion, however, ignores the pres- ence of chlorine, and is, therefore, inexact to some extent. We are not dealing with ideall}^ pure compounds. HEULANDITE. Pure, white heulandite from Berufiord, Iceland, was the material taken for investigation. Upon boiling with sodium carbonate, 1.73 per cent of silica went into solution. From previously ignited heu- landite, only 1.14 per cent was extracted. No silica, therefore, was liberated upon ignition, and a hydrous metasilicate formula for the mineral seems to be improbable. Only one lot of the ammonium 32 ACTION OF AMMONIUM CHLORIDE ON SILICATES. [bull. 207. chloride derivative was prepared, and its composition, together with that of the henlandite, is given below. Heulandite. Ammoninm salt. SiO.^ rfi. 10 16.82 .07 6.95 .46 1.25 .43 } } 61.24 AI2O3 MgO CaO - ----- 18. 00 2.56 SrO Na^O - .--. K2O .60 NH3 - - 4.42 H2O at 100° . - - - -- 3.61 13.00 H2O above 100° 13.63 - 99. 68 100. 45 Here, again, we have the same kind of transformation as before, but rather less complete than in the case of stilbite. That the ammo- nium taken up is equivalent to the bases removed is shown by a study of the ratios. Ignoring water and the soluble silica, the heulandilc ratios are as follows : ^ 48-'^ ISO-^-'-SSO^^ 923^2495? or, uniting bases, I^"l54Al33o(Si308)goo(Si04)24. Again simplifying, this becomes ^ 154-^^330-^3245 or very nearly 1:2:2, as in stilbite. Similarly discussed, the ammonium salt gives the ratios K 27QUa4gAlgg2blio2iOa74g, equivalent to ' ^'362^1353X353, or 1 : 1 : 1. In both cases the orthosilicate molecules are few, and the com- pounds approximate to trisilicates very closely. CHABAZITE. Characteristic flesh-colored crystals from Wassons Bluif, Nova Scotia. The analysis and fractional water determinations are — SiOa 50.78 AI2O3 .-__---._ 17.18 Fe^Os .40 MgO .04 CaO 7.84 Na^O 1.28 • K,0 . .73 H,0 . 21.85 100. 10 '"'^s^iGElf''] CHABAZITE. 33 Fractional water. At 100° 5.22 At 180° :_. 5.70 At 250° 3. 92 At 350° 2. 36 Low redness 4. 51 Full redness .13 Over blast .01 21.85 The unignited mineral, upon boiling with sodium carbonate, gave 0.86 per cent of soluble silica. After ignition only 0.53 per cent was soluble. Here again no silica is liberated b.y calcination, and metasili- cate formulae may be disregarded. Two samples of the ammonium chloride derivative were prepared, which after thorough washing were dried at 40° to 50°. As in the case of stilbite, small quantities of chlorine appear in the compound, not removable by washing. The amount of change effected is also somewhat less than with stilbite, and about the same as with heuland- ite. The analyses of the two samples are subjoined, with the remain- ing alkali all reckoned as soda: A. B. SiO, AlA -.... CaO Na,0(K,0) NH3 H2O CI Less O 55.88 56.09 19.15 19.49 2.25 2.01 .35 .24 4.64 4.83 16.57 16.01 .95 1.35 99.79 100. 02 .21 .30 ).58 99.72 In B, 1.50 per cent of soluble silica was found. After ignition this was reduced to 1.12 per cent. No liberation of silica accompanies the splitting off of water and ammonia. Upon studying the molecular ratios for chabazite and its derivative, relations appear precisely like those found for stilbite and heulandite. For chabazite itself, rejecting water and the 0.86 per cent of soluble silica, we have ^ 58^%4lAl34o!5l832v)2344J or, consolidating soda with lime, Cai7oAl34o(Si308)246(S104)94. 9506— Ko. ^07—02- 34 ACTION OF AMMONIUM CHLORIDE ON SILICATES. [bull. 207. One step further and this becomes Cai7oAl34QX34o=l : 2: 2. Treating derivative "B " m tlie same way, and ignoring chlorine 9S an unexplained impurity, the analysis gives (NH4)284Na8Ca35Al383(Si308)266(Si04)ii2; or, consolidating bases as before, ^^'362^82X378=1 : 1 : 1 nearly. The assumption of commingled ortho- and trisilicate molecules con- forms to Streng's theory of the constitution of chabazite. THOMSONITE. The compact-fibrous variety from Table Mountain, near Golden, Colo. Analytical data as follows : SiOa 41.13 AI2O3 29.58 CaO 11.25 NaaO 5.31 H2O ■ 13.13 100.40 Fractional water. At 100° 1.01 At 180° 1.44 At 250° 1-05 At 350° 3.90 Low redness 5. 65 Over blast .08 13.13 Before ignition the mineral yielded 0.45 per cent of silica to sodium carbonate solution. After ignition 0.68 per cent was soluble. The difference is trifling. Two samples of the ammonium chloride derivative were prepared. In A the heating was only to 300°, in B to 350°. Analyses of the leached products gave the following results: A. B. SiOa 42.41 30. 50 10.00 2.63 2.45 11.96 42.65 A1,0, 31.34 CaO . 9.23 NaaO 2.48 NH3 ._ _... 2.67 H2O .- 11.81 99.95 100.18 CLAUKE AND] STEIGER ] TH0M80NITE AND L A UMONTITE. 35 In A, 1.80 per cent of solnble silica was found. In this case tlie amount of change is very much less than witli the zeolites previously examined. Little lime was removed, and only about half of the soda. Both samples were prepared with six hours of heating in the sealed tube, and it seemed to be desirable to deter- mine whether a more prolonged treatment would produce any greater effect. Accordingly a third lot of thomsonite was mixed with ammo- nium chloride and heated in a sealed tube to 350° for twenty-four hours. The leached product contained 3.40 per cent of ammonia, a distinct increase over the other findings, although the amount of trans- formation into an ammonium salt was still only moderate. We have already seen that stilbite, heulandite, and chabazite approximate more or less nearly to trisilicates in their composition. Thomsonite, however, is essentially an orthosilicate, with variable admixtures of trisilicate molecules. In the example under considera- tion, ignoring water and soluble silica, the molecular ratios give this formula : Nai72Ca2oiAl58o(Si308)5o(Si04)g28; or, condensing, ^ 287^1580-^578 = 1 ^ 2 : 2. Here the acid radicles are ten-elevenths orthosilicate. Ammonium derivative A, similarly computed, gives first — (NH4)i4,Na84Cai78Alg98(Si308)4i (8104)55,; or, uniting univalent bases with lime, ^ 292-^-'^598-^595=^l ^ 2 : 2; the fundamental ratios being practically unchanged. It will be observed that in all of these computations of formulae we have assumed that all the water is "zeolitic;" that is, independent of the true chemical molecules. This question, however, needs to be separately investigated for each individual species. While the assumption is valid for some of these minerals, it is not necessarily valid for all. The real chemical differences between the zeolites are 3^et to be determined; our work merely proves that ammonium com- pounds are formed, completely in some cases, i?artially in ethers. The research should be extended to cover all the zeolites; but this task we must leave to other investigators. LAUMONTITE. Upon this species only one rather crude experiment has been tried, and that upon material of unknown origin. The mineral was heated with ammonium chloride in a sealed tube as usual, and then leached with water. 4.51 per cent of lime and 0.35 of soda were extracted, and in the residue 3. 95 per cent of ammonia was found. Laumontite, therefore, behaves much like the other zeolites, and is only partially transformed into an ammonium compound. 36 ACTION OF AMMONIUM CHLORIDE ON SILICATES. [bull. 207. PECTOLITE. The pectolite whicli was chosen for examination was the well-known radiated variety from Bergen Hill, N. J. The mineral was in long white needles, and apparently quite pure, but the analysis shows that it contained some carbonate as an impurity. Enough of the material was ground up to furnish a uniform sample for the entire seiries of experiments, and the work properly began with a complete analysis. The results obtained are as follows : SiO^ 53.34 AI2O3 .33 CaO 33.33 MnO .45 Na^O 9.11 H,0 2.97 CO, - 67 100. 10 Fractional ivater. At 105° - 0.27 At 180° .16 At 300° .22 At redness 2. 32 2.97 All of the water was given of£ at a barely visible red heat, and the figures show that practically all of it is constitutional — a fact which perhaps hardlj^ needed reverification. The analysis gives the accepted formula for pectolite, HNaCagSigOg. Does this I'epresent, as is commonly assumed, a true metasilicate? If it does, we should expect that ignition would split off silica pro- portional to the acid hydrogen, or one-sixth of the total amount. To answer this question several portions of the pectolite were sharply ignited, to complete dehydration, and then boiled each for fifteen minutes with a solution of sodium carbonate containing 250 grams to the liter. In the extract so obtained the silica was determined, and the three experiments gave the following percentages : 8.96 8.67 8.42 Mean, 8.68 One-sixth of the total silica is 8.89 per cent, and the experiments, therefore, justify the original expectation. The belief that pectolite is a metasilicate is effectively confirmed. Upon the unignited pectolite the sodium carbonate solution has a slow decomposing action, both silica and bases being withdrawn. In two experiments fifteen minutes of boiling extracted 2.07 and 2.55 per cent of silica, and by a treatment lasting four days 4.80 per cent CLARtlE ASD"] STEIGER. J PECTOLITE. 37 was taken out. With water alone similar results were obtained, the action being so rapid, although relatively slight, that pectolite, moistened, gives an immediate and deep coloration with phenol phthalein. By boiling the powdered pectolite with distilled water alone, 1.G5 per cent of silica was brought into solution, and the ignited mineral, similarly treated for fifteen minutes, gave 1.78 per cent. The extraction in these cases is really an extraction of alkaline silicate, as the two following experiments prove. In A the unignited pectolite was boiled for fourteen hours with distilled water, and in B the mineral after ignition was subjected to like treatment for four hours. The dissolved matter in each case was determined, with the subjoined results: Extracted. A. B. SiOa . 2.98 ;30 .81 3 03 CaO 10 Na^O 1.50 4.09 4.63 In A no simple ratio appears, but in B the extracted silicate approxi- mates very nearly to the salt NaaSigOg. In each instance the ratios vary widely from those of the original mineral, showing that actual decomposition and not a solution of the pectolite, as such, has occurred. Schneider and Clarke/* in their first experiments upon the ammo- nium chloride reaction, treated pectolite from Bergen Hill three times successively with the reagent and then leached out with water. In the solution 20.50 per cent of lime and 6.95 of soda were found, showing that a very considerable decomposition had taken place, but the residue was not examined. In a preliminary experiment by the sealed tube method we found that 20.72 per cent of lime and 6.46 of soda were taken out, while i.44 per cent of ammonia was retained by the residue. That is, two-thirds of the bases, approxi- mately, had been converted into chlorides by the reaction. The open crucible and the sealed tube gave essentially the same results, although the retention of ammonia was not noticed by Schneider and Clarke. In order to obtain further light upon pectolite we continued our experiments with the sealed tube method, and have obtained very variable results. All of the heatings with ammonium chloride were conducted at 350°, and the pectolite used was from the same Bergen Hill specimen wTiich served us for our previous work. Our data are as follows, including for convenience of comparison the preliminary experiment which was cited above: A. Heated six hours. On leaching, 20.72 per cent of lime, 6.46 soda, and 0.11 alumina dissolved. The residue contained 1.44 per cent of ammonia. a Bull. U. S. Geol. Survey No. 113, p. 34,1893. 38 ACTION OF AMMONIUM CHLORIDE ON SILICATES. [bdll. 207. B. Heated six hours. 20.10 per cent lime and 5.80 of soda extracted. 1.45 per cent ammonia in the residue. The residue was also examined for silica soluble in 25 per cent sodiiim carbonate solution (on fifteen minutes boiling) , and 43.38 per cent was found. C. Heated six hours. Soluble portion neglected. The residue contained 2.23 per cent of ammonia and 61.79 per cent of soluble silica. The full analysis of this residue is given later. D. Heated ten hours. A complex breaking up of the pectolite took place, and leaching with water extracted the following percentages: SiOa 5.43 AI2O3 .22 CaO 28.20 MnO .23 Na^O 8.29 The residue from this leaching contained 39.63 of soluble silica, but ammonia was not determined. These results are so irregular that definite conclusions can hardly be drawn from them. A and B agree fairlj^ with each other, and also with the earlier work of Schneider and Clarke. C contains more ammonia, but differs widely from B as to the amount of soluble silica in the residue. D, which represents a long heating, indicates a more complete reaction than was observed in either of the other cases. An ammonium compound, however, is evidently formed during the reaction, although its precise nature can not be determined from the evidence now in hand. Something may be inferred from the follow- ing figures, which are to be summarized thus: First, we reproduce from our earlier paper the analysis of the pectolite itself. Secondly, we give the analysis of the insoluble residue obtained in experiment C. The third column of figures is obtained by subtracting from the second column 61.79 of soluble silica and 1.18 of hygroscopic water, and recalculating the remainder to 100 per cent. The fourth column contains the molecular ratios calculated from the third. Pectolite. Residue found. Residue reduced. Ratios. SiOa - - 53.34 .33 33.23 .45 9.11 75.98 .08 9.56 .24 1.84 2.23 1.18 9.47 37.74 .19 25.43 .63 4.89 5.93 0.629 A1,0, . -- -- .002 CaO .454 MnO .009 Na^O NH3 .079 .349 H2O at 100° .27 2.70 .67 H^O above 100° 25.19 1.399 CO, - - 100. 10 100. 58 100. 00 CLARKE AND STEIGER ] WOLLASTONITE AND APOPHYLLITE. 39 These ratios roughly suggest the formation of a salt approximating in composition to the formula R'aCaaSisOa-GlIaO, in which R' is about two-thirds ammonium and one-third sodium. The large amount of water found was doubtless absorbed during the process of leaching. Pectolite itself has the formula NaHCaaSigOg, so that the existence of a hydrous ammonium pectolite is indicated; a conclusion which is probable but not proved. The reaction between pectolite and ammo- nium chloride is possiblj^ simple at first, but followed by or entangled with secondarj?- changes which obscure the results. The experiments are interesting, however, as showing how widely pectolite differs from the other minerals which we have studied, as regards the ammonium chloride reaction. WOLLASTONITE. The only data relative to the action of ammonium chloride upon wollastonite are those given in the original paper by Schneider and Clarke, but on account of the close relationship between this species and pectolite it seems desirable to reproduce the record here. The mineral studied was from Diana. N. Y., and it had the subjoined composition : SiOa 50.05 AlA^FeA 1.13 CaO 47.10 NagO .--- undet. MgO .09 H^O --. .45 98. 82 After two heatings with ammonium cliloride in an open crucible, 36.98 per cent of lime became soluble in water. In other words, a very notable decomposition had occurred, as in the case of pectolite. Since wollastonite is an anhydrous mineral, this result shows that the reaction does not depend upon the presence of hydroxyl. APOPHYLLITE. Upon this species only one rather crude experiment was made, and that with material of unknown locality. Heated with ammonium chloride in a sealed tube, it gave up, on leaching with water, 21.59 per cent of lime and 5.18 of potassa. The residue contained only 0.79 per cent of ammonia. Evidently the mineral, like pectolite and wollastonite, is largely decomposed by the reagent; but it is uncertain whether any regular ammonium compound is formed. It must be remembered that apophyllite sometimes contains small quantities of ammonia, and hence it seems that a more complete investigation of it is desirable. -v 40 ACTION OF AMMONIUM CHLORIDE ON SILICATES. [bull. 207. DATOLITE. The compact, porcelain-like datolite from Lake Superior. This was heated in a sealed tube with ammonium chloride in the usual way. After leaching the product with water, the washed residue con- tained 91.09 per cent of silica and 1.17 of ammonia. Evidently the datolite molecule had been thoroughly broken down, with nearly complete removal of the bases and the boric acid. The significance of the retained ammonia, however, is not clear. EL^OLITE. On account of their interest as rock-forming minerals, the three species nephelite var. elaeolite, sodalite, and cancrinite were studied consecutively and with some reference to one another. The elseolite was the characteristic material from the elseolite-syenite of Litchfield, Me., and had the following composition: SiOa 45.91 AI2O3 31.14 FeaOs .34 FeO . .23 CaO .33 Na.fi 14.60 K2O 5.60 H^Oat 100° .47 H2O above 100° .93 CO2 .40 99. 95 Five grams of mineral were thoroughly mixed with 20 grams of ammonium chloride by long grinding in an agate mortar, and then heated for six hours in a sealed tube to 350°. Even during the grind- ing a strong smell of ammonia was noticeable, and upon opening the sealed tube after heating, a slight pressure of ammonia gas was observed. On extraction with water the following bases passed into solution : FeA-Al203 0.29 CaO . .07 Alkalies (calculated as soda) 2. 10 The residue from the leach water was dried at 50°, and then found to contain 0.92 per cent of ammonia. These figures confirm those obtained in a much less careful preliminary experiment, and show that elaeolite is but slightly affected by the reagent. CT.ATIKE AND"! STEIGER. J CANCRINITE. 41 CANCRINITE. The material studied was the well-known bright yellow cancrinite from Litchfield, Me., and an analysis of it gave the following results: SiO^ 36.19 AI2O3 29.24 Fe^fis trace CaO 4.72 Na^O 19.20 K2O .14 H2O. 4.15 CO2 - - 6.11 99.75 Upon boiling the powdered mineral for fifteen minutes with the standard solution of sodium carbonate, 0.55 per cent of silica went into solution. After ignition, only 0.32 per cent was soluble. No silica, therefore, had been split off by heating. With ammonium chloride two experiments were made. In each case the mineral was intimately ground with four times its weight of the chloride, and heated to 350° in a sealed tube for four hours. During grinding a strong smell of ammonia was noticed, and still more was given off when the tubes were opened. The products were leached with water, and the thoroughly washed residues were ana- lyzed, as follows : SiO. AlA CaO Na,0(+K20) . NH3 HjOatlOO"..- H2O above 100 CO. 99.85 B. 37.48 37.51 31.23 31.98 5.10 5.30 7.78 7. 53 4.73 3.77 1.29 12.24 } 14.48 none none 100. 57 In the wash water from product B, 11.73 per cent of the original soda was found, with no lime, and 0.16 per cent of silica and alumina. Somewhat less than two-thirds of the soda had been taken out. The lime seems to be much more stably combined, and water was taken up, probably in the process of leaching. The carbonic acid of the cancrinite had been completely eliminated. 42 ACTION OF AMMONIUM CHLOEIDE ON" SILICATES. [bull. 207. Apparently, if the product of tlie reaction is a definite compound, the effect of the ammonium chloride has been to transform the cancri- nite into a zeolitic body, approximating roughly to the general formula RiAlSiO^. H2O, but with a small excess of the univalent bases. Analysis A, adjusted by rejecting the 1.29 per cent of hygroscopic water, and recalculation of the remainder to 100 per cent, assumes the following form and gives the appended ratios: Analysis re- duced. Ratios. SiOs -- 38.03 31.69 5.17 7.89 4.80 12.42 0.634 AI2O3 --- CaO - - --- .311 .093 NaaO - - - . .127 NH3 : .282 H2O - J .690 100.00 The substance is evidently not absolutely pure, a condition which might have been expected. Any closer attempt at precise formula- tion would therefore be useless. It most nearly resembles, among the products which we have obtained, the ammonium derivative of thomsonite. SODALITE. Dark-blue sodalite from Kicking Horse Pass, British Columbia. Analysis as follows: SiO^ 39.66 AlA 30.09 Fe^Og .31 CaO -.. . .18 Na^O 22.60 K,0 1.14 H^OatlOO" .17 H,0 above 100" .79 CI 6.12 101.06 LessO=Cl 1.39 99.67 With ammonium chloride two preparations were made, both by the sealed-tube method at 350°. In A the heating lasted twenty-four hours; and in B six hours. From residue A, by leaching with water, 2.96 per cent of alkali, reckoned as soda, was extracted; and from B, CLARKE AND STEIGER ] SODALITE AND THE FELDSPARS. 43 3.53 per cent. In the washed residues the following determinations were made, but complete analysis seemed to be unnecessary. A. B. SiO^ A1..0,(Fe,Oo) 39.33 31.40 .20 20.86 .45 5.92 40.00 32. 34 CaO Na.,0(K20) ... NH3 _- .72 CI - - - _ -- Evidently the amount of change was slight, and no definite ammo- nium derivative had been formed. In one way these results shed some light upon the constitution of sodalite. According to Lemberg and his pupils the mineral is a double salt, a molecular compound of sodium chloride with a silicate like nepheline. If this view were correct sodium and chlorine should be removed together by the action of a decomposing reagent. We find, however, that about 3 per cent of soda was removed from sodalite in forming residue A, while practically all of the chlorine remains behind. So far, then, the evidence is adverse to the view just cited and favorable to that of Brogger, which assigns the mineral, as an atomic compound, to a place in the garnet group. On the other hand, sodium chloride may be volatilized from sodalite by prolonged heating. Two portions of the mineral were each heated for four hours over a blast-lamp flame, losing 10.80 and 10.72 per cent, respectivelj^ The chlorine in the mineral, 6.12 per cent, corresponds to 10.08 per cent of NaCl; to this must be added the 0.91 of water found, making a total possible loss of 11.01 per cent. In the residue from the first lot ignited 0.20 of chlorine was found, so that the vola- tilization of sodium chloride had been almost complete. This reac- tion, however, taking place at a very high temperature, may be only a result of metathesis, and not by any means a proof that sodium chloride, as such, is an essential constituent of sodalite. The evi- dence derived from the ammonium chloride reaction is entitled to the greater weight. THE FELDSPARS. The results which we have obtained with these important rock- forming minerals are interesting only in so far as they show a trifling sensitiveness on the part of the several species toward dissociating ammonium chloride. The action upon them is slight, and ammonium derivatives do not seem to be formed. The data may be briefly sum- marized as follows : Ortlioclase. — From southeastern Pennsjdvania, exact locality unknown. Quite pure cleavage masses. Heated for six hours with 44 ACTION OF AMMONIUM CHLORIDE ON SILICATES. [bull. 207. ammonium chloride to 350° in a sealed tube, and leached with water, 1.52 per cent of KCl went into solution. The residue, dried at 50°, contained 0.20 per cent of ammonia. Oligoclase. — The transparent variety from Bakersville, N. C. Treated like the orthoclase. In the leach water 0.96 per cent of lime and 2.71 of soda were found. The air-dried residue contained 1.47 per cent of ammonia. It is barely possible that in this case an ammo- nium derivative may have been produced, but the data are not posi- tive enough to warrant any definite conclusion. Albite. — Well-crystallized and very pure material from Amelia Courthouse, Ya. Treated like the two preceding feldspars. Upon leaching, 0.12 percent of lime and 0.84 of soda went into solution. In the residue, dried at 50°, 0.32 per cent of ammonia was retained. OLIVINE. Green, transparent pebbles from near Fort Wingate, N". Mex. Examined by Schneider and Clarke, who employed only the open crucible method. By treatment with ammonium chloride only 0.44 per cent of magnesia was rendered soluble in water — i. e., converted into magnesium chloride. In view of the ready solubility of this mineral in even weak aqueous acids, this lack of sensitiveness to ammonium chloride is somewhat remarkable. ILVAITE. This rare mineral was found by Mr. Waldemar Lindgren at the Golconda mine, South Mountain, Owyhee County, Idaho. It occurs in jet black masses and occasional rough crystals, embedded in quartz or calcite, and intimately associated with two other minerals which appear to be garnet and tremolite. Traces of pyrite also appear. The specific gravity of the ilvaite, as determined by Dr. Hillebrand, is 4.059 at 31°. Upon grinding the powdered mineral with ammonium chloride in an agate mortar, a distinct smell of ammonia was noticeable. Three tubes of the mixture were heated to 350°, and one exploded because of the liberation of gas within. Upon opening the second and third tubes, a strong outrush of ammonia was observed. When the con- tents of these tubes were leached with water, large quantities of ferrous chloride went into solution, which, rapidly oxidizing, formed a deposit of brownish hydroxide, and interfered seriously with filtra- tion. The greater part of the lime in the ilvaite was dissolved also. The washed residue, containing much ferric hydroxide, was partially analyzed, and enough' data were obtained to show that a general breaking down of the ilvaite molecule had been effected. Apparently, also, small quantities of an ammonium derivative had been formed; CLARKE and! STEIGER. J ILVAITE AND RIEBECKITE. 45 but this point is uncertain. The original rainei-al was analyzed by Dr. W. F. Hillebrand, and his analysis, contrasted with that of the leached residue, is here given: Ilvaite (Hille- brand). Residue (Steiger). ■ SiO,... - 39.16 .52 30.40 39.14 5.15 13.02 .15 .08 43. 01 Al^Os 40. 08 FeA FeO 8.75 MnO .85 CaO 2.25 MgO undet. Na20 .--. Tindet. NH3 .88 H20at 105°--- .-'- .15 2.64 undet. H2O above 105° undet. CI (a) 100. 41 95. 82 a Small amount. In the leached residue from the third tube 21.37 percent of soluble silica was found — silica which had been liberated during the reaction between the ilvaite and the ammonium chloride. In short, ilvaite behaves toward the reagent much like pectolite, and the product is a mixture of uncertain character. The evident instability of the ilvaite molecule may account for its rarity as a mineral species. Only exceptional conditions would favor its formation. RIEBECKITE (?). The results obtained with ilvaite made it desirable to study, for comparison, some other silicates of iron. Among these the mineral from St. Peters Dome, near Pikes Peak, Colorado, originally described by Koenig as arfvedsonite, but identified by Lacroix as near riebeck- ite, happened to be available. It was treated with ammonium chlo- ride in the usual way and no presence of liberated gas was noticed when the tube was opened. On leaching the product with water, fer- rous chloride went into solution and ferric hydroxide with some manganic hydroxide was deposited. In the leached mass 6.90 per cent of soluble silica was found, and in the wash water from the leaching there was 6.76 per cent of soda. According to Koenig's analj'sis the mineral contains 8.33 percent of soda, so that a large portion of the total amount had been extracted. There was also, 46 ACTION OF AMMONIUM CHLOEIDE ON SILICi^TES. [bull. 207. evidently, a considerable breaking down of the molecule, but no definite ammonium derivative had been formed. This is shown by the following analysis of the leached residue, which is contrasted with Koenig's published analysis'^ of the original mineral in order to indicate the amount of change. In the third column of figures we give the amount of each constituent which could be dissolved out from the residue by treatment with hydrochloric acid. Riebeckite (Koenig). Residue (Steiger). Soluble portion. SiOj-- 49.83 1.43 .75 14.87 18.86 1.75 .41 67. 54 TiO, ZrOs FeoO, 21.28 4.94 .64 none trace I 1.04 .53 3.33 trace 15.74 FeO 4.94 MnO : .64 MgO -_ __. CaO NaaO --. 8.33 1.44 K2O NH3 _ .53 H2O .20 CI 97.87 99.30 The residue is evidently a mixture of free silica and ferric hydrate with probably at least two silicates, one soluble, the other insoluble in hydrochloric acid. The reaction itself is noteworthy because of the fact that the original mineral is but slightly attacked when boiled with strong hydrochloric acid. The other minerals so far studied by us are all easily decomposable by acids, while this one is quite refrac- tory. The energetic character of the ammonium chloride reaction is thus strongly emphasized. iEGIRITE. Material from the well-known locality at Magnet Cove, Arkansas. Not absolutely pure, but somewhat contaminated by ferric hydroxide. This impurity is evident in a discussion of the ratios furnished by the analysis, but is not serious. It does not affect the problems under consideration. By heating with ammonium chloride the mineral was only slightly changed. In the leach water from the product there 'Dana's System of Mineralogy, 6tli ed., p. 400. CLARKE AND 1 STEIGER. J ^GIRITE AND CALAMINE. 47 were l.<3