UC-NRLF ANNALS OF THE NEW YORK ACADEMY OF SCIENCES Vol. XXII, pp. 161-224 Editor, EDMUND OTIS HOVEY METAMORPHISM OF PORTLAND CEMENT BY ALBERT B. P ACINI NEW YORK PUBLISHED BY THE ACADEMY 10 SEPTEMBER, 1912 X N sO THE NEW YOEK ACADEMY OF SCIENCES (LYCEUM OF NATURAL HISTORY, 1817-1876) OFFICERS, 1912 President — EMERSON McMiLLiN, 40 Wall Street Vice-Presidents — J. EDMUND WOODMAN, FREDERIC A. LUCAS, CHARLES LANE POOR, E. S. WOODWORTH Corresponding Secretary — HENRY E. CRAMPTON, American Museum Recording Secretary — EDMUND OTIS HOVEY, American Museum treasurer — HENRY L. DOHERTY, 60 Wall Street Librarian — EALPH W. TOWER, American Museum Editor — EDMUND OTIS HOVEY, American Museum SECTION OF GEOLOGY AND MINERALOGY Chairman — J. E. WOODMAN, N. Y. University Secretary — CHARLES P. BERKEY Columbia University SECTION OF BIOLOGY Chairman — FREDERIC A. LUCAS, American Museum Secretary — WILLIAM K. GREGORY, American Museum SECTION OF ASTRONOMY, PHYSICS AND CHEMISTRY Chairman — CHARLES LANE POOR, Columbia University Secretary — F. M. PEDERSEN, College of the City of New York SECTION OF ANTHROPOLOGY AND PSYCHOLOGY Chairman — K. S. WOODWORTH, Columbia University Secretary— FREDERIC LYMAN WELLS, Columbia University The sessions of the Academy are held on Monday evenings at 8:15 o'clock from October to May, inclusive, at the American Museum of Natural History, 77th Street and Central Park, West. [ANNALS N. Y. ACAD. Sci., Vol. XXII, pp. 101-224. 10 September, 1912] METAMOEPHISM OF PORTLAND CEMENT l % BY ALBERT B. PACINI (Read before the Academy, Part I on 8 January, 1912; Part II, 1 April, 1912) CONTENTS PART I rage Introduction 102 Nature of the problem 163" Chemical composition 104 Mineralogical constitution 165 Setting process 165 \ Hardening process 166 Influence of water upon metainorphisin 168 Temperature of the water at first added "169* Temperature of the water that may subsequently come into contact with the system 171 Quantity of water at first added 172 Size of cement particles 172 Laitance 1 7:> Hydrolysis theory 1 74 Mechanical agitation when water is added 174 Total quantity of water at first added 175, Quantity of water that may subsequently come into contact with the system 175" Surface treatments 177 Membranes 177 Mass treatments 177 Quality of water at first added 178 Having material in solution 178 Quality of water that may subsequently come into contact with the system ISO Having material in solution ISO Sea water 180 Alkali and deep rock waters 181 Having material in suspension . 185 1 A thesis submitted in candidacy for the degree of Doctor of Science at New York University, 1912. Acknowledgments are due to Prof. .7. Edmund Woodman and Mr. Raymond B. Earler of the Department of Geology, New York University, and to Engineer Inspector Ernst Jonson, Board of Water Supply, City of New York, for valuable suggestions made dur- ing the preparation of this paper ; also to Mr. Fred H. Parsons, Assistant Engineer, and Messrs. James E. Jay, Charles M. Montgomery and Charles E. Price, Inspectors, of the Board of Water Supply Laboratory, for material assistance during the experimental work. (161) ANNALS NEW YORK ACADEMY OF SCIENCES PART II Page Experimental investigation 184 Temperature of the water at first added 18 1 650 770 +3 9 12-12 44 2 649 872 +3 -1-4 12-12 60 Normal 628 697 0 0 24-24 34 3 530 622 16 11 22-24 Y 1:3 60 Normal 250 350 0 0 12-12 43 1 253 370 +1 4-6 12-11 44 2 287 327 + 15 +7 12-12 60 Normal 228 317 0 0 22-24 34 3 197 234 14 26 22-24 From these results, it is safe to conclude that, aside from the effects of frost, low temperatures are adverse to the development of the hardening process in cement, and that in general this effect is more pronounced in mortars than in neat cement. The adsorption of calcium hydroxide by the complex hydrogel may proceed at a lower rate at lower temperatures; or if this is not so, the primary hydration, of which this hydrogel is the product, may proceed more slowly, and thus less of the hydrogel be produced, — either of which processes will detract from the hydraulic activities of the mass. It would seem from the experiments that the latter is the more satisfactory expla- nation, since the test specimens which were chilled at first and allowed to return to normal temperature show a tendency to return to normal strength at the longer periods, while the general tendency in the series kept constantly in cold water is to fall further off from the normal, indi- cating only a limited available amount of hvclrosrel to undergo the coa^Ti- O •/ */ C_? o C"1 latin g process. PACINI, METAMORPHI8M OF PORTLAND CEMEXT 189 The effect of sudden chilling at a period when a large proportion of the strength is already developed does not show any decided direction, both the positive and negative variations from the normal averaging the same. It may therefore be concluded that, for the temperatures studied, a chill- ing of this kind has no significant effect. An explanation according to the crystallization theory of hardening would fail to fit the facts so satisfactorily. In the specimens that were chilled at first and allowed to return to normal temperature, there should be under this hypothesis a more significant decrease of strength, owing to the formation of small, non-cohesive crystals from the rapid tempera- ture change. The return to normal conditions should not favor so nearly complete a recuperation as has been noted; unless a re-solution of the crystals and recrystallization were supposed, in which case it may be argued that such a process would require an abnormal solubility of small crystals when compared with large. In a normal specimen, re-solution and recrystallization are undoubtedly going on, strengthening the struc- ture, and the large crystals are growing at the expense of the small. If small crystals preponderate at seven days' age, resulting in a weak mass, it is necessary to postulate a comparatively high solubility of the small crystals in order to arrive at a normal strength at 28 days. This, while by no means impossible, is not probable. Turning to the specimens kept continuously in cold water, it would seem that, although the first chilling should show severe effects, as it did, there should not be such a falling off in the rate of hardening, if the crystallization be progressive. It is quite possible, however, that crys- tallization at this temperature is not favored, and that the total number of binding crystals of calcium hydroxide is therefore less than at normal temperatures. QUANTITY OF WATER AT FIBST ADDED Size of cement particles. — Other factors being equal, the amount of cement rendered inert by the action of water is proportional to the per- centage of fine particles. This is an absolute condition and presupposes free access of water to every particle. Xeedless to say, in practice this condition is seldom realized, except approximately in laying concrete under water, or in the careless use of an excess of water in mixing, or in protracted mixing. In the use of a very fine cement, then, if the proper proportion of water is added, the mixing time carefully regulated and proper precau- tions taken in depositing, the influence of texture upon the strength of the mass occasioned bv the action of water is reduced to a small quantity, 190 AXNALS NEW YORK ACADEMY OF SCIENCES by virtue of the greater hydraulic activity of the fine particles, increasing the impermeability, as will be shown, and the confining therefore of the action of the excess water to a narrow zone. The bulk of the cement will be properly hydrated in spite of- the fineness. The investigation of the effect of the size of particles due to the action of water thereon alone is not feasible, because no satisfactory measure of laitance formation, except the strength of the mass, has been devised. The measure of the strength would be unsatisfactory, since the propor- tion of fine particles affects the strength in other ways than through the formation of laitance, as has been pointed out in a previous communica- tion. From a study of the hydraulic properties of reground cement, Spack- mann and Lesley conclude (93) that only the very fine flour in cement, that portion not measured by the present tests using sieves, reacts when gaged with water and gives strength. It is difficult, of course, to draw a sharp dividing line between active and inactive material in cement, al- though it must be admitted that the greater part of the coarse material, even though it be of the same chemical composition as the fine, has little or no cementing value and serves mainly as a filler. Suitable fractional separation of the portion of cement passing the 200 sieve, by air-elutriation or other method, should with careful study be a valuable guide to the most efficient mechanical composition. Experi- ments upon the first method of separation are recorded by Peterson (71), and a scientific method of fractional elutriation using an inactive liquid has been worked out by Thompson (100). Much should be gained by the application and development of these methods. The influence of the size of particles of inert material added to the cement is also of great consequence, and a proper mechanical grading of the sand used in mor- tars is recognized as vital. The presence of clay in this sand, or the addition of clay alone to cement, come under this category, and have occasioned a great deal of discussion (8, 32, 33, 110). A comparison was made of the permeability of 1 :4 mortar of Portland cement, when used in its ordinary condition, and when screened through a number 200 sieve. P AC IS I, METAMORPHI8M OF PORTLAND CEMENT 191 TABLE 3 Permeability of 2-inch Cubes, Age 28 Days, Subjected to 80 Ibs. Pressure Cement Temperature of water. Deg. Fahrenheit Grams of water passing per hour Number of testa Unscreened Screened A 66 68 68 68 68 68 68 68 64 64 64 64 64 64 68 72 68 68 22 25 29 331 27 31 5 6 71 33 2 Trace 81 Trace 2 0 Trace 0 5, 6 6, 6 6, 6 5, 6 5, 6 6, 6 3, 6 5, 6 6, 6 B c D E F G H.. i :::::. A.verao'e 61 13 The marked decrease in permeability resulting from the use of finer cement in mortar demonstrates that in impermeability, as in strength, the finest particles are the most active factors. Mechanical agitation irlien irate r i* added. — Increased working should weaken a cement after a certain maximum point is passed. In order to •establish this point, the effect of prolonged working was investigated. It was necessary to use a mix of fluid consistency, in which, for obvious reasons, the final set would not under normal conditions take place dur- ing the time over which the experiments were extended. Two grouts were employed: one in which cement was mixed with 30 per cent of its weight of water, and one in which an equal weight of water was used. The different tests were run respectively for periods from one minute to five hours, and they were mixed in a motor-driven stirring machine of the type common in chemical laboratories. After the stated period of stirring, the grouts were poured into glass tubes and kept in a damp closet for the twenty-eight-day period. Cylin- ders exactly two diameters high were cut from the specimens and crushed in the compressing machine, two cylinders being crushed for each period, and the average of the cornpressive strengths being recorded. 192 ANNALS NEW YORK ACADEMY OF SCIENCES TABLE 4 T-wenly-eiglit-day Tests of Grouts Mixed for Varying Lengths of Time Duration of mixing Compressive strength, pounds per square inch, average of duplicate te&ts 50 per ct grout 100 per ct. grout 1 lllinilte 5240 5545 5710 5875 6075 4775 3095 4725 4955 4840 4320 4538 15 minutes 30 minutes . 1 hour 2 hours 5 hours . The effect of mechanical agitation, when thus prolonged, is equivalent to that of the use of excess water — the strength of the cement is pro- gressively diminished as the working proceeds. It is noteworthy that the effect is only reached after a certain optimum period is passed. Be- fore this time, increased working increases the strength. We may con- clude that there occurs within this period a process which neutralizes the effect of hydrolysis; and this process is probably the formation of the network which constitutes the setting. As will be seen later, the effect of excess water is to reduce the ulti- mate strength. The effect, then, of mechanical agitation must be to bring more cement into contact with water and, therefore, to increase hydrolysis. This is probably accomplished by stripping off the protective film of gelatinous material which envelops each cement particle when it comes into contact with water, which film regulates the hydration of cement and causes it to proceed in a regular manner. This film being stripped off, the cement is subject to the destructive action of hydrolysis. Where more water is originally present, the destructive action is sooner attained, as will be seen by comparing the 100 per cent grout with the 50 per cent. Evidently, the setting process proceeds best at high concen- trations, when the amount of water is low. This may be so regulated that the setting process will not take place at all, by using a large excess of water and much mechanical agitation, as has been repeatedly observed, by the writer. Setting time of cement in laboratory air and in damp closet. — The standard specifications for setting-time tests call for storing the specimen in the damp closet, whereas the tests as generally conducted in most laboratories are made in the open laboratory air. A series of experi- ments was made, for the purpose of noting the deviation from standard results caused by this departure from the rule. PACIXI, METAMORPHI8M OF PORTLAND CEMENT 193 TABLE 5 Setting Time in Laboratory Air and in Damp Closet Cement Time of set in minutes Laboratory air Damp closet Initial Final Initial Final x 255 120 300 240 240 375 360 420 360 390 300 300 360 285 250 435 450 480 420 450 Y-l . Y-2 Y-3 Y-4 From these results, it will be seen that setting in a relatively dry- atmosphere takes place in a shorter time than in a damp one; also that the setting time is more uniform under conditions of high atmospheric- humidity. At the same temperature, evaporation takes place more rapidly in the former case ; and allowing a cement mix to stand in such a position that evaporation of the mixing water may readily take place is practically equivalent to the use of an insufficient amount of mixing water. Effect of excess of mixing water on strength of concrete. — Concrete is often mixed so wet that, as it is filled into forms to a depth of several feet, the water rises above the concrete and throws out considerable lai- tance from the cement. The ease of mixing and placing very wet con- crete is the constant incentive for its use. This practice, however, is followed by a great deal of deterioration of the concrete in strength. The strength rapidly decreases with the increase in the quantity of water used in mixing. The visible effect of this weakening is the forma- tion of laitance, which has little or no setting power or strength, and which represents the loss of an active part of the cement, since, as is recognized, the finer parts are more hydraulically active. Tests were made by mixing concrete at normal consistency and shovel- ing one-half the batch into a tank containing three to four inches of water, the depth of concrete being about four inches. The water rose to about an equal depth above the concrete. In test Xo. 1, the concrete was allowed to settle in water four inches in depth for 30 minutes, when the excess of water was siphoned off and the remaining material poured into molds. In test Xo. 2, the depth of water in the tank was three inches, and the water was siphoned off immediately while in agitation. In test Xo. 3, the same process was repeated, except that the depth of the water 194 ANNALS NEW YORK ACADEMY OF SCIENCES was four inches, and the concrete used was somewhat leaner. Test No. 4 represents the direct qualitative effect of the addition of an excess quan- tity of mixing water without subsequent handling. All specimens were cylinders six inches in diameter and 12 inches high. The remainder of the batch of concrete in each case was poured directly into molds, and the specimens were broken at 28 days. The amount of cement lost was roughly ascertained where possible by filtering the siphoned water and weighing the amount retained on the filter. TABLE 6 Effect of Excess of Mixing Water on Strength of Concrete Test No. No. of specimens Proportions Per cent of water Strength at 28 days, pounds per sq. in. Per cent of cement lost 1 I5 3 2 1:2:4 1:2:4 8.2 8.2 1240 760 2 9 5 M 3 2 1:2:4 1:2:4 8.2 8.2 1485 770 io 3 35 3 3 1 : 2.33 : 5 1 : 2.33 : 5 8.2 8.2 1490 315 12 4 4 3 3 1:2:4 1:2:4 8.2 10.3 1385 1155 i 5 Specimens shoveled into water as described. Evidently, then, the mere presence of an excess of water is sufficient to produce the weakening effect, independently of any actual removal of cement from the concrete. As may be seen from Nos. 1 to 3, the leaner mixes suffer the greater deterioration in strength. Effect of excess of mixing water on permeability of concrete. — A par- allel series of tests upon the permeability of concrete treated with an excess of water was made, in which the correspondingly numbered speci- mens were treated in the same manner. The cylinders cast from these batches were eight inches in diameter and six inches in length, and were cased in the standard manner for permeability tests. Three specimens were made for each test, and at the age of 28 days were submitted first to 40 pounds pressure for one hour, then to 80 pounds for one hour, without interruption. The flow recorded is in grains passing during the last ten minutes of test. PACINI, METAMORPHISM OF PORTLAND CEMENT 195 TABLE 7 Effect of Excess of Mixing Water on Permeability of Concrete Test No. Proportions Per cent of water Temperature of percolating water Grams passing in last ten minutes 40 pounds 80 pounds 1 I6 2 26 3 4 4 1 2 4 1 2 4 1 2 4 1 2 4 1 2.33 5 1 2.33 5 1 2 4 1 2 4 8.2 8.2 8.2 8.2 8.2 8.2 8.2 10.3 67° F. 0 479 0 212 0 1814 38 26 0 456 0 588 21 Not tested 18 80 58° F. 56° F. 60° F. 67° F. 6 Specimens shoveled into water as described above. In the foregoing experiments, the decrease in strength and water- tightness may be referred to the deteriorating influence of excess water upon the cement (16). It may of course be argued that the more marked effects obtained in series 1, 2 and 3 than in series 4 are due to the method of making the tests; that is, that a considerable proportion of the active cement was actually removed from the body of the concrete by siphoning off the supernatant water with its laitance. Effect of excess of mixing water on the strength of neat cement. — With the idea in mind that the weakening effect was independent of the removal of cement (1), a further series of tests was instituted, using a neat cement of good quality. The cement was poured into a series of glass tubes in which increasing proportions of water had been put, the tests representing a series of grouts mixed respectively with 50, 75, 100, 150, 200 and 500 per cent by weight of aement of water. The tubes were shaken for one hour and then allowed to stand for 28 days. The cement settled into the bottom of the tubes in the order of its coarseness, the fine nebulous laitance settling last as a cheesy white layer of increasing thick- ness, as the percentage of water was higher. This layer was carefully trimmed off in preparing the test specimens. On breaking out the cylinders from the tubes at the end of the test period, it was decided to cut each cylinder into two, each exactly one diameter high, carefully noting the respective position of each in the tube. On submitting these to compression it was seen that the direction of difference between the upper and lower layers was not constant, nor 196 ANNALS NEW YORK ACADEMY OF SCIENCES was the difference a significant one, so that it was considered legitimate to average the strengths. It will be seen by the table below that, even without actual removal of any cement, the formation of laitance has a weakening action upon cement. TABLE 8 Cotitpressive Strength of Grouts Mixed icith Varying Proportions of Water Per cent of water Crushing strength, Ibs. per square inch, average of two tests. Age, 28 days 50 6855 75 5900 100 4500 150 3430 200 2960 500 1810 The effect of excess of mixing water is therefore seen to result in decrease of strength as the water increases. Whether the effect is a permanent one was the next question that presented itself. To settle this point, a new series was undertaken, in which a larger number of differing percentages was introduced, and in which the resulting strength at two periods was determined. The cement was mixed with the stated percentage of water, and worked for two minutes, the drier mixes upon the table in the usual fashion, and the wetter mixes merely poured into the tubes and shaken. Paper mailing tubes were used, 2 inches by 48 inches, treated with molten paraffin and sealed with paraffined corks, so as to be absolutely tight. To obviate the effect of possible leakage, the whole series was stored in damp sand. Cylinders two diameters high* were cut from the specimens at the stated periods, each cylinder being cut as nearly as possible the same dis- tance from the bottom, and care was taken to avoid including any of the soft cheesy top portion, the settled laitance. PACINI, METAMORPHI8M OF PORTLAND CEMENT 197 TABLE 9 Compressive Strength of Grouts Mixed with Vaiying Proportions of Water, Over Extended Period (Each result is the average strength of three specimens.) Compressive strength, pounds per Percentage of square inch Per cent gain in strength over 28 days 3 months 28 days 22 7076 7504 6 25 6174 5402 —13 30 4563 6030 32 38 3992 5059 27 50 2991 5312 77 75 2113 4078 93 100 1609 3544 120 150 1270 2379 87 200 1306 2579 97 500 399 1141 186 It is apparent from these figures that the effect of hydrolysis upon the strength of cement is a reversible one, at least to a certain extent, since the specimens in which an excess of water was used in mixing showed a greater recuperative ability at the longer period than the cement in which the normal amount of mixing water, in this case 22 per cent, was used. Upon inspection, it was observed that the three months' specimens showed in each case much less laitance than the similar 28 days' speci- mens had shown, and it was considered probable that the laitance, in standing, had adsorbed free lime from the remainder of the cement, through the activity of the water permeating the mass, and thus reverted to the original condition of the cement, or an approach thereto. An analysis was accordingly made of laitance scraped off from the top of one of the 500 per cent water specimens and thoroughly washed by decanta- tion. It probably represents a maximum condition in the hydrolysis of cement. TABLE ]0 Analysis of Laitance from 500 per cent Specimen As obtained from specimen Treated with lime water Si02 . . 15.28 15.91 Fe,O, 2 28 2.42 ALO, ' 3.98 5.82 CaO . 26.96 36.67 MgO. . 2.86 1.28 so3 : 6.47 2.72 CO2 H2O etc 42.17 35.18 198 ANNALS NEW YORK ACADEMY OF SCIENCES The normal ratio of silica to lime in unset cement may be considered 1 to 2.82. In this material we find the ratio 1 to 1.76. This indicates a great loss of lime; and it was thought possible, that, by adsorption of lime, this laitance might regain at least a part of its hydraulic proper- ties. Accordingly it was digested for several days with lime water at laboratory temperature, filtered off, carefully washed with distilled water and dried, as was the previous sample, at 100° C. An analysis showed the results tabulated in the second column. The ratio of Si02 to CaO had changed to 1 : 2.30. Besides direct metathetical reactions between the components of ce- ment and the water solution which always surrounds a mass of hardening cement, adsorption of various materials from this solution is unques- tionably always going on. Were the fine particles of cement inert chem- ically, this would still take place, by virtue of the enormous total surface which they must present. Clay, it has been demonstrated, has the prop- erty of adsorbing ions of C03 from solutions of carbonates, and of Cl from solutions of chlorides (10). The laitance then may, by adsorption of calcium hydroxide given off from the cement adjacent to it, recover some of the lime lost by it. Whether the lime adsorbed restores the original status of constitution is of course mere speculation. The trend of the strength tests shows that this is probably not so, but that the adsorption is not entirely a reversion of the hydrolytic reaction ; in other words, that "drowned" cement will probably never recover and attain to the strength it would have had with proper hydration. Effect of the presence of clay and dissolved substances. — It is apparent that if the decreased strength be directly referable to the action of the excess water upon the cement, any means of preventing the access of excess water should prevent, if only to a degree, the destructive action. The colloidal nature of clay (6) has been utilized in the water-proofing of concrete, the principle of its action being the formation of continuous gelatinous films throughout the structure, which prevent the passage of water. Although the same problem is not presented in a grout that exists in finished concrete, it is probable that some blanketing action might occur upon the addition of clay to the mixed mass. The point was investigated. To correct for the effect of absorption of part of the mixing water by the admixed clay, a consistency test was made upon a sample of cement to which 10 per cent of clay had been added, and it was found to require 4 per cent more water than the same cement used neat. The clay mixes were accordingly gaged with 4 per cent more water P ACINI, METAMORPHISM OF PORTLAND CEMENT 199 than the corresponding neat cement mixes, and the following series of compressive tests was made : TABLE 11 Effect of Clay upon Destructive Action of Eacess of Mixing Water (Average of two tests at 28 days) Neat cement Cement, 10 per cent of which was replaced by a fat clay (dried) Water, per cent Compressive strength, pounds per square inch Water, per cent Compressive strength, pounds per square inch 50 5782 54 1282 75 3134 79 1328 100 2273 104 2577 150 1896 154 2156 200 1381 204 1320 500 514 504 No strength developed If the action of saline solutions upon cement is to accelerate the hy- drolysis of the latter, it would appear that the destructive action of excess water would be accelerated by the presence therein of saline substances in solution; also, it is legitimate to expect that the addition of clay restraining the hydrolysis due to excess water will in this case exert a. similar influence. The following experiments, parallel to the foregoing ones, elaborate this point: TABLE 12 Effect of Clay upon accelerated destructive Action of Mixing Water Containing 5 per cent of Magnesium Sulphate (Average of two tests at 28 days) Neat cement Cement, 10 per cent of which was replaced by a fat clay (dried) 5 per cent solution of magnesium Compressive strength, pounds 5 percent solution of magnesium Compressive strength, pounds sulphate, per square sulphate, per square per cent inch per cent inch 50 2196 54 2774 75 548 79 1608 100 1512 104 No strength 150 556 154 200 No strength 204 « « 500 No strength 504 a a 200 ANNALS NEW YORK ACADEMY OF SCIENCES From these two series of experiments, it is qualitatively apparent that the presence of clay does prevent a certain amount of hydrolysis. From the first series, it is seen that this effect only begins to show itself as higher percentages of water are present, which would indicate that the clay may have taken up much more water than the constituency test revealed, and that, in the relatively drier mixes with clay, the cement suffered in strength because of insufficient water. On the other hand, experiments at this laboratory in which clay was used, replacing up to 10 per cent of cement in normally gaged material, showed that no signifi- cant decrease in strength was thereby obtained ; hence the loss in strength in the 54 and 79 per cent grouts cannot be due to this cause. It is more probable that the colloidal nature of the added clay is brought into play more effectively at the concentrations in which in- creased strength is observed, and that the latter is due to the coagulation of the clay by electrolytes adsorbed at this optimum concentration. The same result would obtain w^here additional saline material has been added to the mixing water, as in the series where a 5 per cent solu- tion of magnesium sulphate was used. The clay here prevents the accel- eration of hydrolysis by the magnesium sulphate through adsorption of part thereof, and possibly by coagulating, forming an impenetrable bar- rier to the further action of water upon the remainder of the cement. QUANTITY OF WATER THAT MAY SUBSEQUENTLY COME INTO CONTACT WITH THE SYSTEM Permeability. — The solvent effect of water coming into contact with cement structures is best studied by the permeability test. This consists in forcing water through a mortar or concrete at a known pressure and observing the amount of leakage through the specimen. In detail, the specimen is generally made up in the form of a cylinder, and this is cased with a thick coating of neat cement on all sides but the bottom. The water, under pressure, is applied on the full cross-section of the specimen and forced through, dripping from the bottom, whence it may be collected. With neat cement, of course, this method is inapplicable, because of the density of the material and the consequently enormous pressure nec- essary to force water through it, and moreover because of the mechanical difficulty in confining the water strictly to a passage through the speci- men. The specimens tested, then, are lean mortars and concretes. Although this test is designed to ascertain the resistance which these materials offer to the flow of water, it is evident that this resistance is not a constant quantity in the case under consideration. PACINI, METAMORPHISM OF PORTLAND CEMENT 201 The temperature and pressure of the percolating water being constant, the flow is diminished by cementing and clogging, and increased by ero- sion and solution; the quantity of water flowing through the mortar or concrete therefore is a function of the balancing of these processes. Cementing may result from deposition of material originally in solu- tion in the percolating water, or dissolved from one portion of the structure and deposited in another. Clogging, similarly, results from material originally in suspension in the percolating water, and deposited in the pores of the concrete, or from material eroded from one part of the mass, either mechanically or as a result of solution of the attacking portions, and deposited in another part. Erosion per se is a negligible factor; that is, the flow of pure water, carrying no suspended matter, will have very small mechanical effect upon an insoluble material. When the water is armed with suspended matter, however, its corrasive effects become proportionally magnified. Solution is the most important factor in the process of percolation. Following the order laid down by Van Hise for natural rocks (104, p. 536), the basic materials removed are, firstly, the alkalies and, secondly, the alkaline earths, in the order calcium, magnesium. Since the alkalies exist in cement in the proportion of a little over one per cent and are not essential to the hydraulic properties or the strength, their solution is a matter of little consequence, except in that it may result in the for- mation of solutions which react upon the lime compounds and render their solution more easy of accomplishment. This reaction has been considered elsewhere. The removal of magnesium compounds proceeds at a lesser rate, although there is a greater percentage of them present; and their removal, in the main, may be dismissed as insignificant. Since more than half the weight of fresh cement consists of lime, and since the strength of cement depends for the greater part upon calcium hydroxide, whether crystalline or adsorbed by colloids, the removal of calcium hydroxide from set cement is the factor of the greatest impor- tance. Considering its solubility in pure water, the reversion of the hydroxide to the crystalline form tends to diminish its solubility, or from the other standpoint, its adsorption by a colloid tends to remove it from the solvent action of water. Unfortunately, however, it must be borne in mind that without exception, cement structures are nowhere subject to the action of pure water alone. From rain water, with its appreciable burden of dissolved gases and atmospheric salts, to the water of the ocean and the more heavily laden rock and mine waters, concrete structures are everywhere in contact with saline solutions of varying concentrations. 202 ANNALS NEW YORK ACADEMY OF SCIENCES The effect of solution in percolation, then, is to a small degree de- pendent upon the solubility of the components in pure water. This effect diminishes as time goes on, because of the reversion of the soluble ma- terial to a less soluble form and because of the protection afforded by the insoluble portions of the system decreasing the exposed area of soluble material. The washing away of these protecting films will of course neutralize the second factor. The increased solubility of the components of set cement in solutions of various electrolytes is the more important element in percolation. Even a very dilute solution may have tremen- dous total solvent power, when the time element is considered. In fact, it may be that the action of a dilute solution will on the whole exceed that of a concentrated solution, by reason of the greater cementing and choking action of the latter, tending to diminish the quantity of water that may come into contact with the soluble portions. A dilute solution, therefore, with its more insidious attack, is probably more to be feared in the end than the strong brine. Observation of the behavior of concretes and mortars during the per- meability tests gives a clue to the balancing of these processes, whether there is a preponderance of cementing and clogging on the one hand, or of solution and erosion on the other. Attempts were made, in the experi- ments noted below, to study chemically the reactions involved, by peri- odical analyses of the percolating water. To this end nearly four hun- dred complete analyses of the effluent water were made. Upon tabulation of these it was observed that any deductions based upon them would be inconclusive, as the chemical composition of the effluent water repre- sented one of a great number of variable factors that might occur at any point either within or without the concrete. The single qualitative generalization, that lime was removed from the cement at a diminishing rate, is the only permissible conclusion from the analytical data. The original purpose of these tests was to ascertain the suitability of various aggregates for use in concrete, with reference to their stability in the presence of percolating water. At the conclusion of the series, it was found that the effect of water upon the various aggregates was prac- tically negligible, during the period of observation, and that the action had been confined to the cement of the mortar. The aggregates had been protected from the action of water by the cement, it being probable, how- ever, that a continuation of the tests would have revealed the action of water upon these rocks, when the protective influence was removed. A series of sixteen aggregates was used, in as many concrete specimens. Since it is not the purpose of this report to discuss the relative suita- bility of these materials for concrete construction, but only to consider PACINI, METAMORPHISM OF PORTLAND CEMENT 203 the action of the water upon the cement, two cases alone will be con- sidered. The rock was crushed and screened for each experiment to the same average effective size, corresponding to the following mechanical analysis : TABLE 13 Mechanical Analysis of Aggregate used in Permeability Tests Sieve Square mesh opening, in inches Per cent passing 1% 1.89 100 W 1.58 94 1 1.02 59 % .78 32 2A .59 21 2 .48 16 3 .30 6 4 .22 0 The sieve ratings are based on diameters of spheres of equivalent vol- ume to the largest sized stone particles that will pass. The fine aggregate was crushed quartz, the standard sand formerly used for cement testing, passing the Xo. 20 and retained on the Xo. 30 sieve. The cement used was a standard Portland of high quality. The specimens were made in the laboratory's standard form for per- meability test, cylinders eight inches in diameter and six inches in length, the proportions used being 1 : 3.5 : 6, this being found the richest mix practicable to secure the porosity required for the test. They were cased in neat cement, and connected suitably for subjection to the pressure of the city's water mains. Each specimen was protected from the direct flow of the water by a layer of one inch of clean coarse sand. The average pressure for the period of observation (52 weeks) was 22 pounds. The determinations of the rate of leakage were made weekly at first, and later every two weeks until the end of the test. The data appended below represent observations on the rate of percola- tion of water through two of the specimens which present the greatest interest from the standpoint of this paper, this flow being recorded in grams passing in ten minutes. The aggregate used in one specimen was a hardened neat cement, crushed to the size stated, and used in place of the rock generally employed in concrete. The parallel specimen selected for comparison was one in which the aggregate was a crushed granite, which showed a low solubility in hydrochloric acid (2.66 per cent dis- solved in one hour's treatment with 1 : 1 HC1). 204 ANNALS NEW YORK ACADEMY OF SCIENCES Temperature records of the percolating water were not kept, since these tests represent a part of a larger series in which this would have been impracticable. The other aggregates tested showed results from which it was quite difficult to draw any legitimate conclusions as to the relative : suitability of different rocks in concrete subjected to these conditions. Concretes containing different aggregates. — A scries of tests on con- cretes made up of different aggregates but with the same cements gave results which may be tabulated as follows : TABLE 14 in Grams of Water passing in 10 Minutes through Concrete Specimens subjected to continuous Water Pressure for 52 Weel-s Grams passing in 10 minutes Time Pressure, pounds per square inch Month Concrete with aggregate of crushed hardened neat cement Concrete with aggregate ot crushed granite 24 hours 25 January — 2111 60 1 week 22 February . . 836 31 2 weeks 22 662 26 3 20 626 40 4 25 570 60 5 22 March 603 62 6 20 530 50 7 20 1295 38 8 25 1127 40 10 25 April 1310 45 12 26 870 25 14 24 May 997 36 16 24 985 28 18 20 June 973 32 20 17 639 20 22 20 792 40 24 17 July 731 36 26 22 802 43 28 26 August 800 49 30 22 78 1 46 32 20 September. 763 50 34 19 115 Trace 36 26 October. . . . 105 2 38 26 107 2 40 23 November. . 110 3 42 25 93 2 44 22 December. . 75 5 46 21 70 3 48 25 80 10 50 20 January. . . . 73 7 52 20 78 7 PACINI, METAMORPHISM OF PORTLAND CEMENT 205 Comparison of these two sets of figures indicates that the cement of the concrete is more attacked than the aggregate. In fact,, the flow obtained in this specimen was the highest but one of a series of sixteen, and the total lime content of the effluent water was also the highest but one. The visible effect upon examination of the interior of the specimens was a bleaching of the mortar, with evident solution of the cement. The original percentage of lime in the mortars was 12.8. Analysis of mortar from the granite specimen showed a content of 4.8 per cent, indicating that nearly two-thirds of the lime had been dissolved out. Further evi- dence of the loss of lime was found in the heavy white crust which formed on the exposed bottoms of the concrete specimens during the test. Small stalactites, quite soft to the touch, were abundant. The quantity of this deposit was not visibly different in the different tests. The calculated loss in lime of the mortar was greater than the loss computed from periodical chemical analyses of the effluent water, and this is due to the fact that much of the dissolved lime was deposited upon the bottoms of the specimens as the stalactitic growth above mentioned. There was no evidence that suspended impurities in the water had been carried into the interior of the concrete, and it is therefore supposed that the one-inch layer of sand by which the latter was screened from the direct flow of the water was an efficient filter for the purpose. The clogging action resulting from this source may therefore be dismissed as negligible. It may be concluded from these tests that concrete of this density tends to protect itself automatically from the action of percolating water, so that, for the period investigated at least, the flow tends to diminish to a minimum. The action of the water seems to be confined to the cement of the mortar, leaving the aggregate relatively unaffected. It is evident that, notwithstanding the utmost precaution in mixing concrete test specimens, wide differences in permeability niay obtain in specimens mixed under the same conditions of handling and by the same workman, owing to structural differences in the resulting mass. How- ever, the results obtained are fairly comparable. The most sensitive test for the internal changes which the concrete has undergone during percolation is the resulting strength of the concrete. Concretes containing different cements. — A series of tests was under- taken in which the specimens were made up in the same proportions, 1 : 2.5 : 6, using in each specimen the same coarse aggregate, a crushed granite, and the same fine aggregate, a standard quartz, but using differ- ent brands of -cement. The specimens were stored in damp sand for a period of 28 days, then subjected to continuous water pressure of about 206 ANNALS NEW YORK ACADEMY OF SCIENCES 25 pounds for a period of 11 months. Parallel specimens were stored in damp sand during this period and allowed to attain their full normal strength. The table following shows the leakage and final strength of the specimens: TABLE 15 Percolation through Concrete Specimens Months of percolation Brands of cement and grams of water passing in 10 minutes A B c D E F K- 146 155 56 37 72 71 68 57 40 286 125 70 47 28 12 28 46 43 7 13 63 22 90 52 37 31 34 14 12 'io 10 164 179 167 161 65 15 11 6 2 5 1 76 16 11 11 7 17 26 16 5 'i 2 230 82 85 82 45 39 33 21 11 14 19 1. . . 2 3 . . . . 4 5 6 7 8 9 10 13 8 11 TABLE 16 Comparison of Strength before and after Permeability Test A B c D E F Compressive strength of specimens at the end of period 7707 490 640 890 750 590 Compressive strength of untreated specimens, pounds per square inch.. Loss of strength through percolation (per cent) 1080 99 1210 60 1230 48 1125 21 1220 39 1090 46 7 One specimen crushed. Other results are average of two specimens. Effect of the direction of flow through concrete. — Concrete seems to offer less resistance to the flow of water when the direction of the flow is parallel to the bed than when at right angles to it. A test covering this point was made with 8-inch cubes of concrete of the proportions 1:4:14, fine and coarse aggregate being a standard crushed Milestone. PACIXI, METAMORPHI8M OF PORTLAND CEMENT 20' TABLE 17 Rate of Flow in Gallons per Square Foot per Hour under 20-ineli Head Age of specimens, 67 days. Temperature of water, 64° F. In specimen parallel to bed In specimen perpendicular to bed 1st 2 minutes 740.96 585.28 636.31 535.53 549.10 mmersed 24 hours, then r 665.38 642.77 662.80 641.15 659.57 164.14 159.54 163.49 158.33 157.93 tested : 182.46 177.54 177.54 177.06 173.67 2d " 3d " 4th " 5th " Specimens i 1st 2 minutes 2d " 3d 4th " 5th In denser concretes, this effect was not found so marked. It will be noted that after storage following the first exposure to the effect of per- colating water, these specimens appear to offer less resistance to the flow of water. This may be due to the fact that in lean concretes the propor- tion of capillary and subcapillary voids is smaller and that of super- capillary voids greater, and that cementing and clogging actions, which have their greatest effect in capillary and subcapillary passages, are not so effective. The greater flow along the bedding planes has been observed in the •case of rock, and is in all respects a phenomenon of the same nature. In the case of a stratified sandstone cited by King (51), the reason is ad- vanced that no more water can pass the more open layers, when advancing -across the bedding planes, than was able to pass those of the closest tex- ture ; whereas when the flow is along the bedding planes, each particular .stratum carries water in proportion to the coarseness of its texture, uninfluenced by any other. In the case of water percolating into a concrete tunnel this would tend to emphasize lateral percolation, and in the case of disintegration would exercise, in general, a localizing influence. It is not to be as- sumed that this is a rigid rule, inasmuch as a large number of factors, evidently, may neutralize this influence. From these considerations, it will be seen that the solvent effect of water upon set cement is of high importance in considering the perma- nence of concrete structures, and that this solvent effect tends to diminish .as the set cement ages. This is not the onlv wav, of course, that water 208 ANNALS NEW YORK ACADEMY OF SCIENCES may afterwards affect the metamorphism of cement. It has been pointed out by Goldbeck (43) and by White (108) that the expansion' or con- traction of concrete depends upon whether the concrete remains wet or dry, and that the strains caused by alternate wetting and drying of con- crete may be a more fruitful cause of cracks than temperature changes. The presence of an optimum quantity of water is necessary, however, so that the proper reactions take place in the mass of setting cement, in order that the strength may increase normally. QUALITY OF WATER AT FIRST ADDED Cornpressive strengths of neat cements gaged ivitli various solutions. — A normal Portland cement was mixed with the proper quantity of water (21 per cent by weight) in which was dissolved, in the different tests, varying concentrations of the salts indicated in the subjoined table. The cement was worked for one minute, and the plastic mass was tamped into glass cylinders approximately one inch in diameter, with the utmost precaution to avoid all air bubbles and at the same time to subject all specimens to the same pressure. TABLE 18 Compressi'veJStrenfftTis of Neat Cement Mixed iritli Solutions of Various Salts (Age of specimens, 28 days. Average of two determinations) Mixes Pounds per square inch Gain or loss, per cent 1 Distilled water 7330 2. 25% rock water8 diluted with distilled water 3. 50% do. 4. 75% do. 5 Rock water alone 6340 6495 6870 5605 -14 -11 - 6 —23 6 2% sodium chloride solution 6675 g 7. 4% do 5815 -21 8 6% do 5065 — 31 9. 8% do. 4215 —43 10 10% do 5285 —29 11. Saturated solution of calcium sulphate (± 0.2% ) . 12. 0 '2% solution of calcium chloride 7025 6960 4 13 02% solution of magnesium sulphate 6680 — 9 14. 0.2% solution of magnesium chloride 15. Equal parts of 11 and 12 (CaSO4 and CaCl2) 16. Equal parts of 13 and 14 (MgSO4 and MgClj 17. Equal parts of 12 and 13 (Ca012 and MgSO4) 18. Equal parts of 11 and 14 ( CaSO4 and MgCl2) 5595 6565 7355 5810 6200 -23 -10 +0.6 -21 —15 8 This water contained: CaO. 1177 parts per million. MgO, 226 S03, 408 Cl, 4360 PACINI, METAMORPHISM OF PORTLAND CEMENT 209 The glass cylinders containing the cement were then stored in a damp closet for 28 days,, when the cylinders were broken out, and two speci- mens, each exactly one diameter high, cut from each cylinder. These were put into water for a few hours, so that they might be in the moist state when crushed. The cylinders were kept in the damp closet instead of being stored under water, to avoid leaching out the salts contained in the mixing water, thus obtaining the maximum effect of the dissolved salts. It will be noted that there is a decided loss of strength in all but one case (number 16). This particular case may be explained by the prob- able formation of an oxychloride, by the magnesium chloride and the magnesium hydroxide liberated by the action of the magnesium sulphate upon the calcium hydroxide of the cement. The oxychloride formed from these two materials has a tensile strength far superior To that of Portland cement itself, and its presence probably counteracted the de- structive action of the salts upon the cement. It is probable, however, that, at longer periods, this increase would disappear and become a de- crease. Otherwise, the presence of saline matter dissolved in the mixing water seems to have a decided deleterious effect upon the strength of cement. This point is of marked importance in construction, inasmuch as the problem of mixing water is often solved by using the water nearest at hand, without inquiry into its qualities. It is the custom to specify that the water used in mixing concrete shall be free from oil, acid, strong alkalies or vegetable matter (77) ; but such a specification does not cover the case in point, and the presence of large quantities of dissolved salts in water used for construction is easily over- looked. In concrete construction, it is of the utmost importance that the water which may be used in mixing be additionally subjected to such tests as will reveal either its mineral content or its action when mixed with cement and possible subsequent attack thereon. The action of sodium chloride appears to be nearly directly propor- tional to the amount employed. This salt is used in mixing water for construction carried on in cold weather, in order to prevent freezing of the deposited concrete. Its effect upon the strength of cement, if used in excessive quantities, is, as has been shown above, likely to become a seri- ous matter. Under the conditions of construction which generally pre- vail, however, much of the salt may be leached out of the mass. The results above represent a condition of maximum attack. Dieckmann (25) recommends the use of from 1 to 2.5 per cent of salt for concrete to be laid in cold weather, but states that percentages larger than this cause a marked decrease in the strength. 210 ANNALS NEW YORK ACADEMY OF SCIENCES Effect of gaging with various solutions upon the strength of mortars afterward stored in water. — The above tests do not show, of course, a normal condition, since no water came into contact with the cement after it had set. Working with more porous material, a 1 : 3 mortar, so that in storage a heightened subsequent water action might take place, the following results were obtained : TABLE 10 Effect of various Salts dissolved in tlie Mixing Water, upon tlie Strength of 1:3 Mortar (Sand, screened Cow Bay. Specimens stored in damp closet for 24 hours, then continuously in water for the rest of period) Mixes Compressive strength, pounds per square inch Number of specimens 7 days 28 days 3 months Water 815 1005 945 1010 885 910 865 935 1090 930 840 1105 1000 1475 1185 1310 1520 1240 1625 1410 1595 1580 1605 1490 1385 1035 2600 1805 2170 2420 2100 2145 2115 2710 2500 2670 2710 2000 1685 1,3,3 3,2,1 3,3,3 3,3,3 1,3,3 2,3,3 2,3,3 3,3,2 3,1,2 3,3,3 2,1,1 3,3,3 2,3,3 5,6,6 6,6,6 4,6,5 6,6,6 4,6,6 3,6,6 4,6,6 2,5,6 3,6,6 5,6,6 5,6,6 6,6,6 5,6,5 I % solution of Ala(S04), 2% do. .... 1 % solution of Na2S04 2% do. 1 % solution of MgS04 1% do. 1 % solution of ZnS04 2% do. 1 % solution of FeSO4 2% do. 1 % solution of NaCl 2% do. Tensile strength, pounds per square inch 7 days 28 days 3 months Water.... 179 208 193 216 205 194 185 126 205 201 184 211 224 272 272 262 290 300 260 246 263 272 266 258 261 279 326 321 340 354 343 317 283 315 319 314 311 310 310 1 % solution of A12 (SO4)3 2% do. 1 % solution of Na.2S04 2% do. 1 % solution of MgS04 2% do. 1 % solution of ZnSO4 2% do. 1 % solution of FeSO4 2% do. 1 % solution of NaCl 2 % do P ACINI, METAMORPHI8M OF PORTLAND CEMENT 211 The general conclusion that may be drawn from these values is that the effect of electrolytes in the mixing water, when the cement is after- wards subject to immersion in water, is to increase the strength at the early periods (7 and 28 days), but later to depress it (15). In general, the more concentrated solutions give a greater depression of strength. The early increase in strength is probably due, in the presence of an optimum quantity of water, to additional cementing or void-filling ma- terial precipitated in the pores of the mortar by reaction between the added electrolytes and the solutions resulting from the action of water upon cement. This deposited material may, in its later history, revert to a soluble form and be washed away, leaving abnormal voids, or else in its growth may disrupt the cells it occupies, in either case reducing the strength. Effect of gaging grout with rock ivaters. — In grouting deep tunnels, the question has arisen as to the advisability of using the rock wrater at hand when fresh water was inaccessible. The water available in the instance in hand was an effluent from a shale bearing a small proportion of pyrites, and when it issued from the rock face it contained a quantity of dissolved hydrogen sulphide. As none of the water was immediately available for a laboratory test, an artificial mixture was made up, in which the quantities of dissolved salts and hydrogen sulphide occurring in the natural water was purposely exaggerated, to obtain accelerated effects. TABLE 20 Analysis of the Artificial!!/ Mineralized Water Parts per million H,S '. 891 CaO 1764 MgO 1461 SO3 1948 Cl 2920 A grout was made up according to specifications, using a normal Port- land cement, and Cow Bay sand with 100 per cent passing 10 sieve, 75 per cent passing 40 sieve; in the proportions 1: iy2 with 35 per cent of liquid. The wet mix was poured into glass cylinders, kept 24 hours in air until set had developed and immersed in water. Four sets of three specimens each were made, the first set mixed with 35 per cent of distilled water; the second, 35 per cent of the water above mentioned; the third, 35 per cent of a 10 per cent dilution of this water, and the fourth, 35 per cent of a 1 per cent dilution. Xo discrepancy was observed in the setting time, as all the specimens 212 ANNALS NEW YORK ACADEMY OF SCIENCES developed a fair set within 24 hours. The grouts mixed with the undi- luted sulphide water turned a dark green, but otherwise no change was noticed in these or any other specimens. Three cylinders one diameter high were cut from each set of specimens, and, after storing 28 days in distilled water, were crushed. TABLE 21 Conipressive Strength of Grout Mixed icith Different Proportions of Water Containing Hydrogen Sulphide (Average of three specimens, age 28 days) Pounds pei- Mixes square inch Distilled water 1424 Undiluted sulphide water 1608 10 per cent of sulphide water, 99 per cent of distilled water. . . . 2088 1 per cent of sulphide water, 99 per cent of distilled water 1110 Apparently, considering the average of the last three values, water of this composition will have no evil effect at 28 days upon the grout with which it is gaged. Three series of tests were undertaken, in which a 1 : 3 mortar of Ot- tawa sand and a cement of good quality was mixed with Croton water, and with two typical rock waters encountered in tunnel work. TABLE 22 Analyses of Rock Waters Parts per million E W CaO 85 943 MgO 159 156 SOS ' 73 172 Cl 1380 3420 Total solids . 2978 7929 The normal amount of water was used to mix the mortars in each case, and the briquettes were stored in the damp closet over the stated periods. TABLE 23 Tensile Strength of 1:3 Mortars Mixed irith Various Saline Waters Pounds per square inch Number of specimens in average 7 days 28 days 3 months Croton water. 302 297 296 322 343 335 344 363 383 6,5,6 6,6,6 6,6,6 Water E Water W PACINI, METAMORPHIS1I OF PORTLAND CEMENT 2lo As was found in the case of the grouts last mentioned, waters of this general concentration do not appear to affect the strength of cement mortars with which they are gaged, and the probabilities are that no serious effects will result from this cause alone. QUALITY OF WATER THAT MAY SUBSEQUENTLY COME INTO CONTACT WITH THE SYSTEM Theoretical considerations. — The action of dissolved salts in water that comes into contact with concrete, where such action is deleterious to the concrete, has been carefully studied by a large number of investi- gators (68, 81, 96, 112). Of the salts which have been found injurious, magnesium sulphate and magnesium chloride seem to have the greatest effects. What concentration of dissolved salts is necessary in order that disintegrating effects shall manifest themselves cannot be definitely stated. This is a field problem and is subject to wide variations under different conditions. A water containing relatively little dissolved material, acting under favorable conditions of porosity, pressure and wide temperature changes upon one concrete, may accomplish failure of the structure; while another water, of high saline content, meeting a dense, impervious con- crete, not forced through the mass by pressure and under conditions of small temperature change, may have practically no action. Manifestly, unless these varying conditions are taken into account, it is unscientific to draw any conclusions regarding the attack of different waters or the resistivity of different cements. It may be laid down as a basic principle, however, that the denser a concrete, other conditions being equal, the greater its resistance to the attack of saline waters (10, 41, 57). The alkali waters of the Western states have given a great deal of trouble in concrete construction. Most experimenters conclude that their action upon concrete is in the main mechanical and due to the disruptive force of crystallizing or efflorescing salts deposited in the pores by intermittent submergence and drying out (30, 38, 49,56). Of course, as has been pointed out, action of this sort is not confined to concrete, and any material of construction possessing porosity is liable to a similar disintegration. The remedy, therefore, is to prevent the penetration of the saline solutions by the employment of courses of permanent, impenetrable materials, preferably asphaltic layers. Where the attack is not mechanical but chemical, this remedy is also applicable. Unfortunately, there are examples of construction which are exceptions, and, in these, some change in the chemical or mechanical 214 ANNALS NEW YORK ACADEMY OF SCIENCES constitution of the cement is the only way to prevent decomposition. In concrete block construction, where the blocks may be made long before they are actually put into the structure, it is found of great advantage to allow them to harden in air or in damp sand, and so permit to a great extent the carbonation of the lime compounds. Some investigators claim excellent results from this method (41, 55). As to the modifications in the constitution of the cement that will combat the action of saline solutions, there is a great disparity of opin- ion, which possibly is based upon lack of standardization of experimental conditions. It is generally conceded that high silica cements are best suited for the purpose (7). The use of puzzolan cements, or of addi- tions of puzzolan to the cement in use, is also well recommended (7, 37, 66) ; and the addition of clay, burnt or dehydrated, finds favor with some (7, 75). As to the lime content of the cement, opinions are divided whether it should be high (5, 41) or low (92). Cement of greater density (57) and cement ground to a greater fine- ness than usual (72) are favorably commented upon. The subject, because of its great complexity and because of the questionable value of laboratory results, is at present in a chaotic state. The length of time that must elapse before judgment may be passed upon the permanence of a material under these conditions and the corresponding newness of the field of Portland cement render present conclusions largely a matter of speculation. Effect of storage in various saline solutions upon the strength of mortar. — In order to study the relative resistance to saline solutions offered by cements varying in chemical composition and in fineness of grinding, a series of 132 2-inch mortar cubes was made up, in the pro- portion of 1:3, with standard Ottawa sand, the cements used being A. A high silica cement B. A low silica cement C. A cement of ordinary composition, sifted and remixed so that 98.8 per cent passed the 100 mesh sieve and 88.6 per cent passed the 200 mesh sieve D. The same cement as C sifted so that 92 per cent passed the 100 sieve and 75 per cent passed the 200 sieve P AC IN I, METAMORPHISM OF PORTLAND CEMENT 215 TABLE 24 Analyses of the Cements Used in Tests with Saline Solutions Per cent A B c SiO2 23.50 19.74 22.99 Fe.Ov . 2.36 2.75 2.42 A12O3 7.28 8.77 6.79 CaO 62 18 60 86 60.84 MgO. 2 29 2.86 4.14 SO3. 1.11 1.39 1.76 C02H2O, alkalies 1.28 3.63 1.06 The cubes were stored 24 hours in the damp closet, and then trans- ferred to the solutions mentioned in the following table, three cubes to each liquid, and there stored for three months, then broken. TABLE 25 Comprcssive Streiif/tli of Mortars Stored for Three Months in Various Saline Solutions (Each value is the average of three determinations) Storage medium Pounds per square inch High silica Gain, pei- cent Low silica Gain, per cent Finely ground Gain, per cent Coarsely ground' Gain, pei- cent 1 Croton water 2217 9462 2134 2066 Sodium 5% 2267 2 2090 -15 3266 53 2273 10 sulphate, 10 % 3264 47 2035 -18 2223 4 2262 9 Magnesium 5% 3244 46 1787 -28 2233 5 2759 33 sulphate, 10$ 2604 18 2646 7 3003 42 2489 20 Sodium 5 % 2365 7 1785 -28 2305 8 2695 30 chloride, 10$ 1778 -20 2019 -18 2968 40 2044 -1 Magnesium 5$ 2331 r •J 1827 -26 2731 33 2305 12 chloride, 10$ 1757 O1 1769 -28 2570 20 2269 10 Calcium 5 % 2653 19 25169 2 2219 4 1808 -13 chloride, 10 %\ 2224 0 1994 -19 2042 -4 2238 8 Average gain (per cent) . . 10 -17 .... 20 12 Average of two determinations. 216 .ANNALS NEW YORK ACADEMY OF SCIENCES The general deductions from these experiments for the period covered are that the high silica cement, notwithstanding its slower rate of har- dening, resists the action of these dissolved salts better than the low silica cement, and the finely ground cement better than the coarsely ground. Moreover, with the concentrations used, the stronger solutions in nearly every case had a more destructive effect upon the strength of the mortar than the weaker. The strengths here obtained by storage in salt solutions are in general decidedly greater than those obtained by storage in fresh water. Ex- amination of the cubes, when removed from the solutions at the end of the test period, revealed under a lens that the exterior was being at- tacked, minute pittings being quite distinct. The strength attained by these specimens may be considered as a re- sultant of the balancing of two effects : the deposition of crystallized or precipitated material in the voids, which by packing the spaces with solids will increase the compressive strength; the creation of additional voids by direct solution or by the disruptive effect of metathetically pro- duced material. It is probable that the disintegrating effect for these concentrations is reached considerably beyond three months' exposure. From the increases in the compressive strength, it is likely that at this period a great deal of crystallization or precipitation has proceeded, overbalancing in the main the disruptive effects. This is a general deduction, and single instances are notable in which the reverse holds good. In the case of the finely ground cement, the density of the mortar made therefrom has prevented the disruptive effect to a greater degree ; and thus the deposition, while not necessarily as much as in the coarser cement mortars, has had a more marked effect in increasing the strength. Effect of storage m rock water upon the strenr/lli of lean cement mortars. — A series of briquettes of 1 : 4 Ottawa sand mortars was made up, using a normal Portland cement of high quality. The mix was made, lean purposely to accelerate whatever disintegrating effect might occur. Batches of the briquettes were stored in bottles in the laboratory for the 7-day and 28-day tests, and additional series were stored in the field, for the longer tests, at stations where the waters in question were encoun- tered. The field series were stored in running water, and the action upon these should be more severe than upon the laboratory specimens stored in still water. In each case a parallel test was made by storing a series in pure drinking water. PACINI, METAMORPHISM OF PORTLAND CEMENT 217 TABLE 26 Tensile Strength of 1:4 Mortars, stored in Rock Water Water Strength, pounds per square inch Specimens in average Stored in laboratory Stored in field 7 days Gain 28 days Gain 3 mos. Gain 6 mos. Gain Drinking.. « \ 5 J 211 220 203 221 +4#' -4# +4$ 297 312 287 288 320 323 313 340 "ijf -2# 6£ 324 247 303 328 12,12,6,6 12,12,6,6 12,12,6,6 12,12,6,6 5$ -3% -3fo -24f, - Z% 1% "B". ... « ^i> > TABLE 27 Analyses of Rock Waters in Previous Experiments Parts per million A B H2S.. 44 20 7 284 124 727 826 949 15 4 5 4 399 87 118 38 353 31 546 270 459 317 SiO2 Fe.203-fALA CaO MgO.... SO, Cl CO2 Alkalies, etc . Total solids 3037 1895 751 The drinking water used to store the blanks contained in neither case more than 100 parts per million of total solids. The most consistent reduction of strength, although a slight one, is observed in the case of water B, a fairly typical sulphato-chloride water according to Clarke's classification (18, p. 190). A strikingly high and sudden reduction occurs at six months in water A, a sulphate water charged with hydrogen sulphide, while water C, a chloride water, shows no marked reduction of the strength, which, however, may be due to a' low salinity. The six-month briquettes stored in water A showed superficially much minute pitting, due to the removal of the sand grains, presumably by solution of the matrix of the cement. Two sections were cut from one of these briquettes, one transverse and one longitudinal, in the hope of discovering whether any replacement of the original material by sul- 218 ANNALS NEW YORK ACADEMY OF SCIENCES phates or sulphides was going on. The microscopic examination did not reveal anything of the sort, the sections being in all respects similar to sections cut from the briquettes stored in drinking water. It was con- cluded therefore that the loss of strength was due to actual removal of material by solution rather than by replacement with material which would cause disintegration through a discrepancy in volume. The legitimate general deduction from these tests is that, over the period of experiment, the effect of these waters is greater in void filling by crystallization or precipitation than in disintegration by solution or disruption. The void-filling material, if of a stable nature and not likely to return into solution, should be in a measure a protection against the further entrance of the saline solutions. It has been mentioned that this prop- erty has been suggested of magnesium hydroxide (70). Probably upon this possibility is based the reported effect of chemically inert fine ma- terials, added to the cement for protection against such destructive action. SUMMARY OF EXPERIMENTAL EESULTS 1. Increase of temperature of the water with which cement is mixed causes acceleration of the set up to a certain maximum temperature, then a retardation. 2. Storage in cold water, without freezing, retards the hardening of neat cement, and that of mortars more. 3. Increase in the proportion of fine particles in a cement decreases the permeability of mortar made therefrom. 4. Mechanical agitation increases the strength of cement up to a cer- tain maximum time; after which, if continued, it reduces it. 5. The setting of cement is accelerated by dryness of the atmosphere. 6. An excess of mixing water progressively reduces the strength of cement. This effect is partly reversive of itself, and the reversion may be increased by additional colloidal material in the original cement. 7. Water percolating through concrete dissolves the lime of the ce- ment chiefly, and this effect tends to neutralize itself by "healing." 8. Percolation through concrete preferably follows the bedding planes. 9. Salts in solution in the mixing water tend to lower the strength of cement. This effect may be neutralized by precipitation in the pores. 10. Storage in saline water affects low silica cements more than it does high silica, and coarsely ground cements more than it does finely ground cements. PACINI, METAMORPHISM OF PORTLAND CEMENT 219 GENERAL CONCLUSIONS In general, the metamorphism of Portland cement represents on an accelerated scale the processes which occur in natural rocks. The accel- eration is of course due to the ease with which water has access to the finely comminuted particles in the initial stages of metamorphism. Many of the minerals found in natural rocks, when ground as finely as, or finer than Portland cement, undergo vastly accelerated reactions in the pres- ence of water; colloidal bodies are thereby produced, and the water is rendered alkaline (18). The end product of prolonged water action on Portland cement bears a striking qualitative similarity to the end product in the kaolinization of feldspars. The same transformations evidently occur in both cases, — the alkalies and the lime are abstracted, and the water and alumina con- tents increased. The exceeding fineness and high adsorptive power of the resulting products are also similar. The action of water on nearly all silicate minerals is, in effect, a repetition of this process. The peculiar adsorptive properties of colloidal bodies render these liable to coagulation. As has been pointed out in preceding pages, much of the cementing material of conglomerates and sandstones, except where calcitic, may have its origin in a similar phenomenon. On a natural scale, the action of water is greatly retarded, because of the larger size of the bodies acted upon, and the consequent paucity of surface upon which water may exert its influence. When Portland ce- ment has properly undergone its initial metamorphism, the setting process being complete and the hardening process in great part so, it approaches the condition of a natural metamorphic rock, and activities towards its further change are katamorphic and vastly slower in their results than the initial changes. The component particles have now become consolidated and the surface offered to the action of water is minimized. Of course, this is truer of neat cement than mortar and truer of mortar than of concrete, these being in the order of increasing porosity. The hypothesis that crystal formation is responsible for the strength of hardened cement is not so complete and satisfactory as the colloidal hypothesis just referred to. In a compact mass, the growth of crystals can hardly be considered anything but an element of weakness. As has been shown by the foregoing results, the effects of varying some of the conditions of the action of water upon cement are best explained by considering the hardening a coagulative process rather than a process of crystallization. 220 ANNALS NEW YORK ACADEMY OF SCIENCES BlBLIOGKAPHY 1. Abstract : Influence of the Proportion of Water on the Compressive Strength of Cement Mortar and Concrete. Cone. Eng., 3, 31G. 1908. 2. AMBRONN, H. : Crystallization and Gel Formation in the Hardening of Cement. Tonind. Zig., 33, 270. 1909. 3. ANDERSON, A. O. : The Incrustation and Absorption of Concrete. Proc. Am. Soc. Test, Materials, July, 1911; reported in Cement Age, 13. 1911. 4. ANDERSON, G. G. : The Effect of Alkali on Concrete. Trans. Am. Soc. C. E., 67, 572. 1910. 5. ANON. : Tests on Concrete in Sea Water. Eng. News, 64, 483. 1910. 6. ASHLEY, H. E. : The Colloid Matter of Clay and its Measurement. Bull. 388, U. S. G. S. 1909. 7. BIED AND VIVIERS : Decomposition of Mortars. (Report.) Proc. Internat. Soc. Test. Materials. 1909. 8. BINCKLEY, G. S. : Experiments on the Coarser Particles in Cements. Eng. Record, 61, 212. 1909. 9. BLOUNT, B. : The Setting of Portland Cement. Eng. Digest, 3, 135. 1908. 10. BROWN, F. W. : Action of Alkali Water on Cement. Cement, 9, 93. 11. BURCHARTZ, H. : Researches upon Mortar and Concrete Frozen and again Thawed Out. Mt. Kgl. Materialpruf, 28, 270. 3910. 12. - — : Properties of German Portland Cements. Cement Age, 10, 250. 1908. 13. BURKE AND PINCKNEY : The Action of Alkali Salts upon Portland Cement. J. Ind. and Eng. Chem, 3, 317. 1911. 14. CAMERON, F. K. : Application of the Theory of Solid Solutions to the Study of Soils. Report No. 04. Div. of Soils, Dept. of Agriculture. 15. CARPENTER, R. C. : Materials which Retard the Activity of Portland Cement. Cement, 8, 21. 10. CHAPMAN, C. M. : Waterproofing with Water. Eng. Record, 62, 707. 1910. 17. CLARKE, E. C. : Trans. Am. Soc. C. E., 14, 158. 1885. 18. CLARKE. F. W. : The Data of Geochemistry. (Second Edition.) Bull. 491, U. S. G. S. mil. 19. COCHRAN, J. : Directions and Suggestions for the Inspection of Concrete Materials. Engineering and Contracting, 37, 115. 1912. 20. Committee of Assn. R. R. Superintendents : The Action of Sea Water on Concrete. (Report.) Cone. Eng., 2, 139. 1907. 21. CUSHMAN, A. S. : The Effect of Water on Rock Powders. U. S. Dept. of Agriculture. Bureau of Chemistry. Bulletin 92. 1905. 22. D'ANS, J. : Do Clay and Cement Adsorb CO3 Ions? Z. Chem. Ind. Kolloide, 6, 38. 1910. 23. DECKER, G. : The Hardening of Hydraulic Cements. Tonind. Ztg., 33, 1493. 1909. 24. DESCH, C. H. : The Chemistry and Testing of Cement. London, 1911. 25. DIECKMANN, G. P. : Use of Concrete in Freezing Weather. Rock Prod- ucts, 9, 52. 1909. 20. DITTER AND JESSER : Thermocliemical Experiments upon Sintered Portland Cement Mixtures. Zentralbl. Hydr. Zement. 1, 71. 1910. 27. DOELTER, C. : Fusions of Calcium Metasilicate and Calcium Aluminum Silicate. Zentralbl. Hydr. Zement. 1, 104. 1910. METAMORPHISM OF PORTLAND CEUEXT 221 28. DONATH, E. : Behavior of Certain Liquids towards Cement and Concrete. Chem. Ind.. 34, 323. 1910. 29. DUNN. W. : Effect of Sewage Gases on Concrete Work. Cem. Eng. News, 22, 434. 1910. 30. Editorial : Effect of Salt Solutions on the Set and Durability of Concrete. Cone. Eng., 3, 318. 1908. 31. - — : Effect of Temperature on the Set of Concrete. Cement Record, 1, 24. 32. — — : What Percentage of Clay is it Safe to Permit in Sand for Cement Mortar? Eng. News, 57, 620. 1907. 33. - — : The Effect of Clay in Cement Mortar. Eng. Record, 55, 703. 1907. 34. - — : The Micrography and Constitution of Cement. Tonind. Ztg., 32, 1681. 1908. 35. - — : Retempered Mortar as a Bonding Material. Cement Age, 13, 97. 1911. 36. FALK, M. S. : Cements, Mortars and Concretes. New York. 1904. 37. FERET, M. R. : Resistance of Cements to Sea Water Increased by Admix- tures of Puzzuolan. Cem. Eng. News, 22, 11. 1910. 38. FISK, A. J. : Effect of Alkali on Cement Mortars. Eng. News. 64, 108. 1910. 39. FULLER, W. B. : The Laws of Proportioning Concrete. Report to the Aqueduct Commissioners. New York. 1905. 40. GADD, W. L. : Composition of a Number of Proprietary Mixtures for Waterproofing Concrete. Cone, and Const. Eng. 3, 54. 1908. 41. GARY AND SCHNEIDER: Behavior of Hydraulic Cement in Sea Water. Mt. Kgl. Materialpruf.. 27, 239. 1909. 42. GILLMORE, Gen. Q. A. : Cements, Limes and Plasters. 43. GOLDBECK, A. T. : Expansion and Contraction of Concrete While Harden- ing. Proc. Am. Soc. Test. Mat., 11, 563, 1911. 44. GREENMAN, R. S. : Practical Tests of Sand and Gravel Proposed for Use in Concrete. Proc. Am. Soc. Test. Mat., 1, 575. 1911. 45. GRIMSLEY, C. P.: The Gypsum of Michigan. Geol. Surv. Mich., 9, (2). 138. 1904. 46. HEADDEN, W. P. : The Destruction of Concrete by Alkali. Expt. Sta. Color. State Ag. Coll.. Bull. 132. 47. HEISER, A. : The Grinding Process and the Setting of Portland Cement. Tonind. Ztg., 34, 936. 1910. 48. JANDA. F. : Theories of the Setting and Hardening of Portland and Rom-in Cements. Oesterr. Z. Berg. Iliittenw.. 56, 432. 1908. 49. JEWETT, J. Y. : Cement and Concrete Work of the U. S. Reclamation Service, with Notes on the Action of Alkali Waters. Proc. Am. Soc. Test. Materials, 8, 480. 1908. 50. KARL : Effect of Sewage on Concrete. Eng. Record, 62, 28. 1910. 51. KING, F. H. : Principles and Conditions of the Movement of Ground Waters. 19th Annual Rep., U. S. G. S., part 2. 1899. 52. LE CHATELIER, H. : Recherches sur les Mortiers Hydrauliques. Paris. 1904. 53. - — : The Decomposition of Cements. Proc. 7th Int. Cong. Applied Chem., London, Sec. 2, 10, 5. 1909. 222 ANNALS NEW YORK ACADEMY OF SCIENCES 54. MACKENZIE, W. B. : The Disintegration of Concrete Exposed to Sea Water between High and Low Tides. Eng. News, 58, 472. 1907. 55. MATTHEWS, E. R. : Action of Sea and Fresh Water on Cement and Con- crete. Eng. Digest, 5, 527. 1909. 56. MATTHEWS AND WATSON : Effect of Freezing and Immersion upon Cement and Cement Mortar. Eng. Digest, 4, 405. 1908. 57. MAYNARD, E. : Iron Ore Cement. Cem. Eng. News, 22, 243, 1910. 58. - — : Solution and Decomposition of Cements. Tonind. Ztg., 33, 1462. 1909. 59. McGEE, W .7: Outlines of Hydrology. Bull. Geol. Soc. Amer., 19, 193. 1907. 60. MCKENNA, C. F. : Hardness of Plasters and Cements, and a Simple Chronographic Apparatus for Recording Set. J. I ml. and Eng. Cheiu., 4, 110. 1912. 61. MEADE, RICHARD K. : Portland Cement. Easton, 1906. 62. - — : The Influence of Fine Grinding upon the Physical Properties of Portland Cement. Cement Age, 7, 161. 1905. 63. - — : The Ultimate Composition of Portland Cement. Cliem. Eng., 10, 183. 1910. 64. MICHAELIS, W. A., SR., AND W. A., JR. : The Hardening Process of Hydrau- lic Cements. (Paper read before Assn. Ger. Port. Cement Mfrs., Ber- lin, 1907.) Cem. Eng. News, separate. 65. - — : The Hardening of Cement Under Water. (Paper read before Assn. Ger. Port. Cement Mfrs., Berlin, 1909.) Cem. Eng. News, separate. 66. MICHAELIS, W. A., SR. : Puzzuolan Mortars in Sea Water. Tonind. Ztg., 33, 1308. 1909. 67. MONTEMARTINI, C. : The Hydration of Cements. L'Industria Chimica, 7, 169. 1908. 68. NEWBERRY. S. B. : Cement for Sea Water Construction. Cement Age, 9, 38, 1907. 69. NEWBERRY, S. B. AND W. B. : Constitution of Hydraulic Cements. J. Soc. Chem. Ind., 16, 887. 1897. 70. O'HARA, J. M. : Action of Sea Water on Portland Cement. Cem. Eng. News, 22, 288. 1910. 71. PETERSON, P. M. : Determination of the Finest Powder in Portland Cement. Tonind. Ztg., 33, 1687. 1909. 72. PIERCE, G. : Destructive Action of Alkali upon Cement. Mining Sci., 63, 130. 1911. 73. PLUMB, R. A. : Waterproofing Concrete. Canadian Cement and Cone. Rev.. 4, 91. 74. POIRSON, L. : Studies of the Chemical Influence of Sea Water upon Port- land Cement. Ciment. Nos. 6, 7. 1910. 75. POTTER, C. J. : Chemical Changes in Portland Cement Concrete, and the Action of Sea Water thereon. Jour. Soc. Chem. Ind., 28, 6. 1909. 76. POULSEN, A. : Long Time Concrete Tests in Sea Water. Eng. News. 64, 3. 1910. 77. Progress Report of the Joint Committee on Concrete and Reinforced Con- crete of the American Society for Testing Materials. : Proceedings, 1909. PACINI, METAMORPHISM OF PORTLAND CEMENT 223 78. READ, E. J. : The Crystalline Products of the Hardening of Portland Cement. J. Soc. Chem. Ind., 29, 735. 1910. (Discussion by Blount.) 79. REIBLING AND REYES : Physical and Chemical Properties of Portland Cement, III. Phillip. Jour. Sci., 6, 207. 1910. 80. RICHARDSON, CLIFFORD: Portland Cement from a Physico-chemical Stand- point. Atlantic City, June, 1904. Also, Eng. News, Aug. 11, 1904, Jan. 26, 1905, Cement 1904-1905, 5, 3, et seq. 81. ROHAN. W. D. : The Effect of Sea Water, Alkali Water, and Sewage on Portland Cement. Eng. Cont, 24, 52. 1905. 82. ROHLAND, P. : The Weathering of Stones and Mortars. Zeits. Chem. Ind. Kolloide, 8, 48. 1911. 83. - — : Influence of Electrolytes on the Setting Time of Cement. Zeits. Angew. Chemie, 16, 622, 1903. Stahl u. Eisen, 28, 1815 1908. 84. - — : Causes for the Changes in Speed of Hydration in Cement. Zeits. Chem. Ind. Kolloide, 8, 251. 1911. 85. SABIN, L. C. : Cement and Concrete. McGraw, 1905. 86. SAGER AND CRAMER: The Hardening of Previously Set Cement. Tonind. Ztg., 32, 1746. 1908. 87. SCHOTT, O. : American Portland Cement Concrete in Sea Water. Chem. Ztg.. 35, 256. 1911. 88. SHEPHERD AND RANKIN : Preliminary Report on the Ternary System CaO-ALO3-SiO2. Jour. Ind. and Eng. Chern., 3, 211. 1911. '89. SLIGHTER, C. S. : Theoretical Investigation of the Motion of Ground Water. 19th Ann. Report, U. S. Geol. Surv., pt. 2, 1899. DO. SMITH, A. ERSKINE : Some Old Cement Tests. Cone. Eng., 4, 124. 1909. "91. SOCIETY SCAND. PORTLAND CEMENT MFRS. : Sea Water Tests of Portland Cements. Cement Age, 10, 184. 1908. -92. SPACKMANN. H. S. : Relation of Lime Content of Cement to Durability of Concrete. Concrete, 11, 52. 1911. 93. SPACKMANN AND LESLEY : The Hydraulic Properties of Reground Cement. Cement Age, 10, 1. 1908. 94. SPALDING, F. P. : Hydraulic Cement. Wiley, 1906. 95. STERN. E. : The Setting and Hardening of Cement. Mt. Kgl. Materialpruf, 28, 173. 1910. 96. STEWARDSON, C. W. : The Effect of Alkali on Cement. Brick, 33, 102. 1910. 97. TAYLOR AND THOMPSON : Concrete, Plain and Reinforced. Wiley, 1905. 98. THATCHER, E. : The Effect of Sea Water upon Portland Cement, Mortar and Concrete, and upon Steel Embedded Therein. Proc. Am. Soc. C. E., 34, (2). 1908. 99. THOMAS, E. : Colloidal Cement Waterproofing. Cement Age, 11, 158. 1909. 100. THOMPSON. G. W. : The Classification of Fine Particles According to Size. Proc. Am. Soc. Test. Mat., 10, 601. 1910. 101. TORNEBOHM (quoted in 80) : The Petrography of Portland Cement. Kong. Int. Verb. Material. 1897. 102. TUTTLE, M. C. : Decomposition of Concrete in Sea Water. Eng. News. 60, 341. 1908. 103. UNGER, C. (Tr. by MICHAELIS) : Fused Portland Cement. Cement Eng. News, 22, 192. 1910. 224: ANNALS NEW YORK ACADEMY OF SCIENCES 104. VAN HISE, C. R. : A Treatise on Metamorphism. U. S. Geol. Surv., Mono- graph 47. 1904. 105. WALLING, W. A. B. : Some Notes on the Setting of Cement. Chem. News, 104, 54. 1911. 106. WESTON, R. S. : Effect of Sewage on Concrete. (Letter.) Eng. Record, 61, 714. 1909. 107. WHITE, A. H. : Disintegration of Fresh Cement Floor Surfaces by the Action of Smoke Gases at Low Temperatures. Pr. Am. Soc. Test. Ma- terials, 9, 530. 1909. 108. WHITE, A. H. : Destruction of Cement Mortars and Concrete through Ex- pansion and Contraction. Proc. Am. Soc. Test. Mat, 11, 531. 1911. 109. WIG, R. J. : The Effect of High Pressure Steam on the Crushing Strength of Portland Cement Mortar and Concrete. Proc. Am. Soc. Test. Mat., 11, 580. 1911. 110. WILLIAMS, I. A. : Influence of Fine Grinding upon the Physical Properties of Portland Cement. Trans. Am. Cer. Soc., 10, 244. 1908. 111. WITHEY, M. O. : A Survey of the Concrete Aggregates of Wisconsin. Con- crete, 12, 55. 1912. 112. WORM SEE, D. : Changes in the Strength of Portland Cement. Cement Ager 9, 254. 1907. PUBLICATIONS OP THE NEW YORK ACADEMY OF SCIENCES (LYCEUM OF NATURAL HISTORY, 1817-1876) The publications of the Academy consist of two series, viz. : (1) The Annals (octavo series), established in 1823, contain the scientific contributions and reports of researches, together with the rec- ords of meetings and similar matter. A volume of the Annals coincides in general with the calendar year and is sold at the uniform price of three dollars per volume. The articles composing the volume are printed separately, each in its own cover, and are distributed in bundles on an average of three per year. The price of the separate articles depends upon their length and the number of illus- trations, and may be learned upon application to the Librarian of the Academy. The author receives his separates as soon as his paper has been printed, the date of issue appearing above the title of each paper. (2) The Memoirs (quarto series), established in 1895, are issued at irregular intervals. It is intended that each volume shall be devoted to monographs relating to some particular department of Science. Volume I is devoted to Astronomical Memoirs, Volume II to Zoological Memoirs, etc. The price is one dollar per part as issued. All publications are sent free to Fellows and Active Members. The Annals are sent to Honorary and Corresponding Members desiring them. Subscriptions and inquiries concerning current and back numbers of any of the publications of the Academy should be addressed to THE LIBRARIAN, New York Academy of Sciences, care of American Museum of Natural History, New York, N. Y. ; UNIVERSITY OF CALIFORNIA LIBRARY