GIFT OF 
 
 ... v.' 
 
 WJBI'IZWIH 
 * -NT 
 
6 1888 
 
 CONCLUSIONS 
 
 ADOPTED BY THE 
 
 FEENCH COMMISSION 
 
 IN REFERENCE TO 
 
 TESTS OF CEMENTS. 
 
 THE 
 
 INFLUENCE OF SEA WATER 
 
 HYDRAULIC MORTARS. 
 
 TRANSLATIONS FROM 
 THE FRENCH AND FROM THE GERMAN BY 
 
 O. M. CARTER, 
 
 CAPT., CORPS OF ENGINEERS, T 
 
 E. A. GIESELER, 
 
 I A N T E N li I N E E II. 
 
 WASHINGTON: 
 
 GOVERNMENT PRINTING OFFICE. 
 
 1897. 
 
CONCLUSIONS 
 
 ADOPTED BY THE 
 
 FKENCH COMMISSION 
 
 IN REFERENCE TO 
 
 TESTS OF CEMENTS. 
 
 THE 
 
 INFLUENCE OF SEA WATER 
 
 ON 
 
 HYDKAULIC MORTARS. 
 
 TRANSLATIONS FROM 
 THE FRENCH AND FROM THE GERMAN BY 
 
 O. M. (CARTER, ^ E. A. GIESELER, 
 
 CAPT., CORPS OF ENGINEERS, U. S. A., TJ. S. ASSISTANT ENGINEER. 
 
 WASHINGTON: 
 GOVERNMENT PRINTING OFFICE. 
 
 1897. 
 

 WAR DEPARTMENT. 
 
 DOCVMEXT No. 36. 
 OFFICE OF THE CHIEF OF ENGINEERS. 
 
OFFICE OF THE CHIEF OF ENGINEERS, 
 
 UNITED STATES ARMY, 
 Washington, D. C., Aprils, 1897. 
 
 SIR : I have the honor to submit, herewith, translations by Capt. 
 O. M. Carter, Corps of Engineers, United States Army, and Mr. 
 E. A. Gieseler, United States Assistant Engineer, of the following- 
 named papers : 
 
 1. Conclusions adopted by the French Commission in reference 
 to Tests of Cements. 
 
 2. The Influence of Sea Water on Hydraulic Mortars. By Dr. 
 William Michaelis. 
 
 3. The Influence of Sea Water on Hydraulic Mortars. By 
 Union of German Portland Cement Manufacturers. 
 
 4. The Influence of Sea Water on Hydraulic Mortars. By E. 
 Candlot. 
 
 These papers contain information of great value to officers of 
 the Corps of Engineers, and I recommend that authority be granted 
 to have them printed at the Government Printing Office (in one 
 volume), and that five hundred copies be obtained for the use of 
 the Engineer Department, upon the usual requisition. 
 Very respectfully, your obedient servant, 
 
 JOHN M. WILSON, 
 Brig. Gen., Chief of Engineers, 
 
 United States Army. 
 Hon. R. A. ALGER, 
 
 Secretary of War. 
 
 [First indorsement.] 
 
 WAR DEPARTMENT, April 9, 1897. 
 Approved. 
 
 By order of the Secretary of War : 
 
 JOHN TWEEDALE, 
 
 Chief Clerk. 
 
 (3) 
 
 371950 
 
CONTENTS. 
 
 Page. 
 
 1. Conclusions adopted by the French Commission in reference to Tests 
 
 of Cements 7 
 
 2. The Influence of Sea Water on Hydraulic Mortars. By Dr. William 
 
 Michaelis 37 
 
 3. The Influence of Sea Water on Hydraulic Mortars. By Union of German. 
 
 Portland Cement Manufacturers 63 
 
 4. The Influence of Sea Water on Hydraulic Mortars. By E. Candlot 69 
 
 (5) 
 
CONCLUSIONS ADOPTED BY THE FRENCH COMMISSION IN 
 REFERENCE TO TESTS OF CEMENTS. 
 
 INTRODUCTION. 
 
 A technical commission having for its object "to formulate uni- 
 form rules to be followed in testing construction materials and to 
 determine the units which should be assumed for a basis of com- 
 parison " was constituted under the direction of the ministry of 
 public works of France by a decree of November 9, 1891. That 
 decree specifically instructed one of the two sections of the com- 
 mission, Section B, to make a study of the questions relating to 
 materials of construction other than metals. At the first meeting, 
 presided over by Inspector General Guillemain (December 28, 1891), 
 a certain number of members were formed into a committee which 
 proceeded to make investigations and experiments and to formulate 
 conclusions. That committee confined itself to a study of the 
 methods of testing masonry materials, namely, cements, limes, poz- 
 zuolanas, sands, and plasters. Those studies were the object of its 
 deliberations during twenty-nine meetings held between the 28th 
 of December, 1891, and the 5th of May, 1893, at the ministry of 
 public works, and most of the questions arising in the course of the 
 discussion were submitted to the study of a subcommittee, then 
 treated in special reports. 
 
 The labors of- the committee were finally communicated to the 
 members of the section, which, at its meeting on May 8, 1893, defi- 
 nitely adopted the conclusions to be submitted to the full com- 
 mission. The present report has for its object to sum up those 
 conclusions in a methodical order and to give as succinctly as 
 possible the considerations which led to their adoption. 
 
 GENERAL CONSIDERATIONS. 
 
 1. CHOICE OF TESTS. 
 
 Several members of the section were of opinion that the end to 
 be obtained was to seek among all of the tests now in use a small 
 number which could be recommended for determining the quality 
 of materials. The section would then be able to formulate some 
 clear, precise rules which could be followed for receiving materials. 
 There would be avoided the definition of a considerable number of 
 tests from which only doubtful information concerning the value 
 of the products could be obtained, and whose very diversity would 
 be opposed to the " unification" which was the object of the labors 
 of the commission. 
 
10 TESTS OF CEMENTS. 
 
 The majority, however, diet 'not think it proper to adopt that 
 view. 
 
 On the one hand it would not be possible in the present state of 
 our knowledge of masonry materials to affirm that such and such 
 tests are necessary and sufficient to characterize in a general man- 
 ner good products, and it should be observed on the other hand 
 that the different tests to which a product is submitted are intended 
 to show whether it possesses properties suitable for the special use 
 to which it is to be put. For certain constructors who wish to 
 make use of a cement for temporary sea works rapidity of set will 
 be of prime importance, and strength at the end of some months 
 comparatively unimportant. For others, on the contrary, who are 
 making conduit pipes, rapidity of set is of no special interest, 
 while ultimate strength is a matter of the utmost importance. 
 
 Those two examples, which can be readily multipled, suffice to 
 show that a certain test, indispensable in one case, becomes useless 
 in another, and that consequently the task of the commission would 
 not be completely fulfilled if it left out any one of the tests now 
 used unless that test fails to furnish any information whatever con- 
 cerning the properties sought. Hence, each observer, according 
 to the conditions which the materials must fulfill, is permitted to 
 make a choice among the tests which he believes it useful to make. 
 Some members have expressed an opinion that a distinction should 
 be made between working tests and laboratory tests, the first being 
 simple, practical, rapid, and sufficient for the ordinary conduct of 
 works ; the others of a more scientific character, requiring more 
 complicated apparatus, more experienced operators, and experiments 
 lasting over a greater length of time. That view, however, was 
 not adopted by the section. To the reasons which led to abstain- 
 ing from expressing a preference for any particular test it should 
 be added that, generally, the methods to be employed can not be 
 simplified except to the detriment of exactitude; that it would be 
 difficult to make one classification of working tests and another of 
 laboratory tests, the personnel and the apparatus varying consid- 
 erably according to locations ; and finally, once a method of test 
 is decided upon there is no reason why it should not be conducted 
 with all of the resources which are at the disposal of the operator 
 in the particular case. 
 
 2. STANDARD TESTS. 
 
 The tests now used relate, some to binding media in a pulverized 
 condition, others to the same materials gauged with water or with 
 water and sand to form pastes or mortars. * 
 
 * The name paste is given generally to products resulting from gauging bind- 
 ing media with water, and the name mortar to products resulting from gauging 
 them with water and sand. 
 
TESTS OF CEMEXTS. 11 
 
 Such tests can have for their aim to seek either, given a certain 
 binding medium, what results can be realized by employing it in 
 such or such conditions, or, given several binding media, which 
 among them will best satisfy the special objects in view. 
 
 Thus, a constructor having at his disposal a cement of a cer- 
 tain quality for the manufacture of the masonry of a work, can 
 make tests for the purpose of determining what mixtures will be 
 the most advantageous, having regard to the conditions under 
 which the mortar will be placed ; or, having at the time of formu- 
 lating specifications a choice between such or such limes or cements, 
 he will require of the manufacturers or he will have made tests 
 capable of showing such qualities of the competitive products as 
 will enable him to select one of them with a perfect knowledge of 
 the reasons for such selection. To satisfy such double object the 
 methods adopted in the different tests should apply to the most 
 general case, that in which any mortar whatever is submitted to 
 test ; but for each test it will be necessary to specify definitely the 
 mixture, the conditions of manufacture, of preservation, and the 
 age of any special mortar tested, in order to obtain comparable re- 
 sults which may serve to characterize the products tested. When 
 operations are conducted on a special mortar the test is called a 
 standard test. 
 
 Suppose, for example, it is desired to indicate the method to be 
 followed in tests of rupture by tension : There will be specified the 
 form of the briquette as well as the arrangement of the apparatus 
 to be employed for determining the resistance of any mortar what- 
 ever, and there will be. added that the standard test will be con- 
 ducted on a mortar having been manufactured and preserved under 
 such or such conditions. It is the results of standard tests that 
 manufacturers will probably be required to publish, when they 
 wish to make known the quality of their products. It is also to 
 the standard test that reference is generally made in specifications 
 for furnishing binding media, because once the rules relative to 
 those tests are adopted in the laboratories, the competitors will be 
 able to determine what conditions must be satisfied in each case as 
 easily as the consumers can verify whether the conditions exacted 
 are fulfilled. For each one of the tests relating to hydraulic mor- 
 tars the special rules governing the standard test will be indicated 
 immediately following the general rules defining the methods 
 recommended by the section. 
 
 3. CLASSIFICATION OF PRODUCTS FROM RESULTS OF TESTS. 
 
 The object of the labors of the commission is clearly defined, as 
 well as by the decree which constituted it as by the discourse de- 
 livered by President Picard on assuming office, and it would be 
 
12 TESTS OF CEMENTS. 
 
 useless to revert to it had not various members of the section, on 
 different occasions, expressed doubts on the subject. In their 
 opinion the labors of the commission would not fulfill completely 
 the expectations of those interested in the manufacture and use of 
 masonry materials if, in connection with the best methods of 
 testing, there were not given results or figures characterizing 
 various products of good quality. The section considers that the 
 commission would exceed its powers if it specified the conditions 
 which should be fulfilled by materials in order to be classed, with 
 respect to their commercial value, as belonging in such or such a 
 category. Although the commission would be in a position to 
 determine those conditions perfectly, that is to say, to draft speci- 
 fications for the acceptance of binding media of various kinds, it 
 is certain that a number of clauses would have to be varied accord- 
 ing to the nature of the works to be executed. For a slow-setting 
 cement, for example, it should be stipulated in certain cases that it 
 should set in less than six hours, although in other cases there 
 would be no reason for the same delay ; a resistance of 20 kilograms 
 per square centimeter at the end of seven days would be necessary 
 for certain work, while for others the same resistance at the end 
 of a month would suffice. Thus, for any well-defined class of 
 products the conditions to be inserted in type specifications could 
 not be given, and it is always to the consumer that finally will fall 
 the task of specifying those conditions, reference being had to the 
 conditions which a product must fulfill and the results which the 
 progress of manufacture permit to be realized. As to the com- 
 mercial value of products, it should be based on numerous elements 
 of quality. If we take the example which has just been cited there is 
 no reason for assuming that a cement which after having hardened 
 slowly has acquired at the end of a relatively long time a great 
 resistance, is of a quality, and consequently of a value, superior or 
 inferior to that of a cement of the same nature which after having 
 hardened more quickly has acquired at the end of an equal lapse 
 of time a less resistance. 
 
 The labors of the commission have therefore been limited for 
 good reasons to a choice of methods of test, without specifying a 
 classification of products from results furnished by a series of tests. 
 
 4. CLASSIFICATION OF HYDRAULIC BINDING MEDIA. 
 
 If the plan of assuming as a basis of classification the quality of 
 hydraulic binding media of the same nature was rejected unhesita- 
 tingly, the question as to whether it was possible to proceed to a 
 classification of those materials was discussed at length. 
 
 Several members of the section maintained that such classifica- 
 tion had been adopted by the conventions of Munich in 1884, and 
 
TESTS OF CEMENTS. 13 
 
 of Dresden in' 1886,* and had been brought to the attention of the 
 International Congress held in Paris in 1889 ; that such a classifica- 
 tion would have the advantage of preventing the sale of the most 
 diverse products under identical names, which would favor con- 
 sumers ; and, moreover, that it is necessary, in order to employ 
 the methods of test recommended by the commission, that it should 
 be known at least on what substances the test should be made. 
 
 The majority of the section has noted the difficulties that would 
 be met immediately, as well from the technical as from the com- 
 mercial point of view, if that attempt were made. From the techni- 
 cal point of view the choice of a satisfactory basis of classification 
 would be a very delicate one. Reliance could not be placed upon 
 the essential properties of the products, since those properties, 
 often common to products of very different natures, have no rela- 
 tion to each other, and none of them can be considered as character- 
 istic. Neither would chemical composition be a certain criterion. 
 There is not yet sufficient information as to the grouping of the 
 elements which constitute hydraulic binding media ; even for those 
 media which present qualities most nearly alike the proportion in 
 
 *(1) Hydraulic limes are products obtained by the calcination of limestones 
 containing more or less clay or silicic acid, and which, sprinkled with water, are 
 slaked entirely or partially into powder. According to local circumstances, the 
 lime is delivered in commerce in the form of lumps, or, hydrated, in the form of 
 powder. 
 
 (2) Roman cements are products obtained by the calcination, below the verge 
 of vitrification, of marl containing much clay. They do not slake when sprinkled 
 with water, and it is necessary to employ mechanical means to reduce them to 
 powder. 
 
 (3) Portland cements are products obtained from the calcination, up to the 
 verge of vitrification, of natural marl, or of artificial mixtures of substances con- 
 taining clay and lime. They are reduced to powder by grinding, and contain at 
 least 1. 7 parts, by weight, of lime for 1 part of material which gives to the lime its 
 hydraulic property. To regulate certain properties of technical importance, 
 there may be added foreign material up to 2 per cent of the weight without this 
 addition necessitating any change of name. 
 
 (4) Hydraulic gangues are natural or artificial materials which generally do 
 not harden under water when alone, but only when mixed with caustic limes. 
 Such are pozzuolana, santorin earth, trass obtained from certain volcanic tufa, 
 furnace slag, burnt clay, etc. 
 
 (5) Pozzuolana cements are products obtained by intimately mixing powdered 
 hydrates of lime with hydraulic gangues, ground to the fineness of dust. 
 
 (6) Mixed cements are products obtained by intimately mixing manufactured 
 cements with suitable extraneous material. Such binding media should be 
 formally designated as mixed cements, with an indication of the materials 
 entering into their composition. 
 
 It was remarked at the congress of 1889 that the preceding nomenclature does 
 not comprise fat or thin limes, makes no distinction between more or less 
 hydraulic limes, does not classify grappier cements obtained as by-products in 
 the manufacture of hydraulic limes, and confounds natural with Portland 
 cements. 
 
14 TESTS OF CEMENTS. 
 
 which, the elements enter varies within very wide limits. Neither 
 can a classification be based upon methods of manufacture. Those 
 methods, for exactly those products which it would be most interest- 
 ing to classify, are of relatively recent invention and are continu- 
 ally being modified ; their definition, insufficient even at present, 
 would become entirely incomplete in the near future. From the 
 commercial point of view a classification would certainly have the 
 result (and the observations exchanged in the committee have 
 abundantly demonstrated this) of changing the name under which 
 certain grades of cement are now sold. Such a change would 
 introduce a great deal of trouble into the manufacture of hydraulic 
 binding media, and would give rise to lively protests. 
 
 For those inconveniences, which are real, will there be found 
 sufficient compensation in the advantages resulting from classify- 
 ing the cements ? It is doubtful. 
 
 A more or less exact definition of the nature of various binding 
 media will not prevent fraud, notably the selling under a certain 
 name of cements or limes which do not possess the corresponding 
 qualities on which the consumer has a right to count. But when- 
 ever there is a real necessity, there should be given the place where 
 the product is obtained, and the tests which the section aims at 
 unifying will permit the completion and control of the information 
 which the manufacturer should have given. 
 
 As to those tests, after having made a detailed study of the sub- 
 ject, the section has recognized that, in fact, it can be proceeded 
 with without the materials tested having been classified into any 
 definite category. 
 
 The methods recommended apply equally well to products of all 
 kinds, such as Portland cements, Roman cements, slag cements, 
 grappier cements, hydraulic limes, etc., and the slight modifications 
 to which those methods should be subjected, so far as the tests 
 relating to rapid-setting cements are concerned, can be followed 
 without the absence of classification causing the least trouble to 
 experimenters. 
 
 The uselessness of classification from the point of view of a choice 
 of methods of tests has exerted a decisive influence on the section, 
 and it has abstained from entering upon a question which could 
 but create in future an obstacle to the adherence of a certain 
 number of interests to the rules promulgated as the result of its 
 studies. 
 
 5. LINE OF WORK ADOPTED BY THE SECTION. 
 
 From the preceding considerations the line of work of the sec- 
 tion has been determined. It occupied itself successively with 
 cements, limes, pozzuolana, santorin earth, and plaster, which are 
 
TESTS OF CEMENTS. 15 
 
 the object of five distinct parts of the present report, the last being 
 devoted to a recapitulation of the conclusions. The methods of 
 testing materials now followed has been examined, compared, and, 
 where necessary, modified with a view of obtaining by the simplest 
 and most practicable processes the most exact information. The 
 section regrets that it has not always arrived at a satisfactory 
 solution, and has been obliged in several cases to point out the 
 defects in the existing methods, expressing the desire that further 
 studies may permit them to be perfected. The methods relating 
 to tests, which, like those of resistance, for example, furnish results 
 varying with the methods and apparatus adopted, have been defined 
 in the most complete detail. The section, however, confines itself 
 to giving simply general instructions whenever it is a question of 
 absolute properties of materials, which, like specific gravity, remain 
 the same whatever method be employed to determine them. In 
 summing the conclusions arising from a study of the numerous 
 questions submitted to its examination, the section has not lost 
 sight of the object to be obtained as defined by President Picard in 
 his discourse of December 28, 1891, "to advise, not to legislate; to 
 indicate methods of experiment, not to insist upon them ; to act by 
 persuasion, by the example of public works, not by coercion. " The 
 section has sought to facilitate the means of arriving at that end 
 by avoiding a disturbance of universally-accepted customs, the 
 injury of French manufactures, or of compromising the possibility 
 of a final understanding with other nations. Although in execu- 
 ting its mission it has exhausted all sources of information, the 
 section is not unaware that its conclusions are far from being com- 
 plete, and that there remains a great deal to be done to decide upon 
 certain and exact methods applicable to every test relating to the 
 study of masonry materials. 
 
 Its efforts, however, will not have been useless if it has been able 
 to assist in laying the foundation of a useful and practical work 
 which can be completed and made perfect in the future. 
 
 RECAPITULATION OF CONCLUSIONS. 
 
 1. GENERAL OBSERVATIONS. 
 
 Although in tests of masonry materials it is almost impossible 
 to determine results with mathematical exactitude, it is neverthe- 
 less useful to take every precaution known in laboratory practice, 
 that new errors may not be grafted on those which are unavoida- 
 ble. Such precautions are too well known to make it necessary to 
 mention them in detail here. Hence the section confines itself to 
 presenting some observations relating to the apparatus to be 
 employed and to the selection and preparation of the test pieces. 
 
16 TESTS OF CEMENTS. 
 
 There lias been given in the course of the detailed report a com- 
 plete description of a certain number of appliances, and in the case 
 of others there have been set forth the special conditions which 
 they should fulfill, having in view the tests for which they are 
 designed. It can be added for the latter that, in a general manner, 
 preference should be given to the simplest instruments, those easi- 
 est to keep in repair, and those which are best adapted to being 
 verified or checked. It is indispensable that every instrument 
 without exception should be verified when it is received, and that 
 those should be reverified later which become worn with use, as, 
 for example, screens and briquette molds, or which have parts sus- 
 ceptible of disarrangement or deterioration, such as balances and 
 certain breaking apparatus. Concerning manner of taking and 
 preparing samples, it must not be forgotten that under the influ- 
 ence of humidity and carbonic acid in the air masonry materials 
 are subject to many changes in their essential properties, some 
 among them becoming profoundly altered. The regulations neces- 
 sary to be adopted, in order to avoid as far as possible the causes 
 of error which might result from a defective manner of taking or 
 preparing samples, should evidently depend 011 the object of the 
 tests, and vary as it is a question, for instance, of determining the 
 quality of cement that a mill is manufacturing, or of deciding upon 
 the receipt of a cargo of cement which is required to satisfy cer- 
 tain known conditions. No general law can be formulated in this 
 respect. There are recapitulated herein the conclusions arrived at 
 by the section following upon a study of the methods recommended 
 for tests of every nature to which masonry materials should be 
 submitted. Referring to the detailed report, the section remarks 
 that those conclusions do not pretend to decide as to the relative 
 value of different tests for determining the qualities of products and 
 the conditions of their reception. 
 
CONCLUSIONS RELATING TO TESTS OF CEMENTS. 
 1. FINENESS OF GRINDING. 
 
 The section recommends the adoption of the following methods 
 for determining the fineness of grinding of cements. 
 
 (a) The sample will be divided into four lots by the aid of three 
 screens with square meshes, defined as follows : 
 
 (1) Screen of 324 meshes per square centimeter, or 18 threads 
 millimeter in diameter per linear centimeter. 
 
 (2) Screen of 900 meshes per square centimeter, or 30 threads 
 millimeter in diameter per linear centimeter. 
 
 (3) Screen of 4,900 meshes per square centimeter, or 70 threads 
 yfo- millimeter in diameter per linear centimeter. 
 
 (b) Tests will be made on a sample of 100 grams. 
 
 (c) Screening by hand will be considered finished when less than 
 jV gram of material will pass through under the action of 25 
 movements of the arm. 
 
 (d) The use of a shaking machine is recommended to eliminate 
 rapidly the greater part of the fine dust. 
 
 (e) Complete screening by machine is also recommended ; but it 
 can not be made the subject of exact regulation, since the condi- 
 tions which the machine should satisfy are not rigorously defined. 
 
 (/) The results for each screen will be expressed by summing 
 the residues which do not pass it. 
 
 2. SPECIFIC GRAVITY. 
 
 As this is a question of an absolute quality of cements, the sec- 
 tion does not believe in recommending one apparatus to the exclu- 
 sion of others, and confines itself to formulating, as follows, the 
 general arrangements applicable to the use of the volumenometers : 
 
 (a) To determine the specific gravity of cements one of the 
 methods actually in use may be employed, provided it permits 
 obtaining the first decimal with certainty and the second within an 
 approximation of two units. 
 
 (6) The precautions to be taken in making the experiments are 
 the following : 
 
 (I) Care will be taken that the cement is freshly pulverized ; the 
 parts retained by a screen of 900 square meshes per square centi- 
 meter, as well as those agglomerated by humidity after they have 
 been reduced to powder and passed through a screen, will be mixed 
 intimately with the rest of the sample, on the total of which the 
 test should be mad.e, 
 
 15671 2 (17) 
 
18 TESTS OF CEMENTS. 
 
 (2) The liquid to be made use of should be benzine or some 
 mineral oil. 
 
 (3) The temperature should remain constant during the entire 
 operation, and should not exceed 15 C. 
 
 3. APPARENT DENSITY. 
 
 The section recommends the adoption of the following rules : 
 
 (a) The apparent density of a cement will be determined by 
 weighing a measure of cylindrical form 10 centimeters high and 
 having a capacity of 1 liter, filled by means of a sieve funnel. 
 
 (b) This apparatus is composed of a vertical funnel whose circu- 
 lar section has a diameter of 2 centimeters at the base and 15 centi- 
 meters at a height of 15 centimeters above the base, at which 
 height is placed a perforated plate having per square decimeter 
 about 1,050 holes 2 millimeters in diameter.* The funnel is pro- 
 longed by a cylindrical spout 2 centimeters in diameter and 10 
 centimeters high. The apparatus is supported by a tripod frame. 
 
 (c) The measure will be placed 5 centimeters below the lower 
 extremity of the spout. The cement will then be poured into the 
 funnel in little masses of from 300 to 400 grams, which will be 
 forced to pass through those screens by stirring with a wooden 
 spatula 4 centimeters wide. The filling will be stopped when the 
 base of the cone, which will rise little by little above the measure, 
 has reached its upper edge. The excess of cement will then be 
 scraped off by moving across the top a straight blade held in a 
 vertical plane. Throughout the entire operation the measure 
 should not be subjected to any trembling or shock. 
 
 (d) There will be adopted as the weight of a liter the mean of the 
 results obtained by five successive measurements. 
 
 (e) It is useful to have the tests made on the cement in the con- 
 dition it is delivered and on the fine dust having passed a screen of 
 4,900 meshes. In all cases the degree of fineness of grinding of the 
 sample used should be indicated, as well as the apparent density. 
 
 4. CHEMICAL ANALYSIS. 
 
 The conclusions that the section proposes to formulate, so far as 
 chemical analysis is concerned, are reduced to the following : 
 
 (a) Operators should be permitted the liberty of using any of 
 the usual methods for determining the chemical composition of 
 cements. 
 
 (b) Complete chemical analyses are recommended ; all of the ele- 
 ments found should be indicated, without grouping, in the record 
 of proceedings of the operation. 
 
 Such plate is found in the market. 
 
TESTS OF CEMENTS. 19 
 
 (e) In default of a complete analysis, the mixture of the volatile 
 materials could furnish useful information. 
 
 5. TESTS OF HOMOGENEITY. 
 
 A study of the information which may be collected from the use 
 of the magnifying glass, in testing cements with respect to homo- 
 geneity, has led the section to the following conclusions : 
 
 (a) The magnifying glass can be employed usefully to give indi- 
 cations of the degree of homogeneity of cements. 
 
 (b) Observations should be made on the material retained by a 
 screen of 4,900 meshes, operating successively with magnifying 
 powers of about 3 diameters for the general examination and of 8 
 for the detailed examination. 
 
 (c) If the examination reveals the presence of grains suspected 
 of coming from foreign materials in the cement, the nature of 
 those can be verified either by complete or by partial chemical 
 analysis of the entire product or of the suspected portions, or by 
 any other means that may be judged most suitable to identify the 
 foreign materials. 
 
 6. MANUFACTURE OF STANDARD PASTES AND MORTARS. 
 
 The section proposes to define, as follows, pastes and mortars on 
 which standard tests should be made : 
 
 1. STANDARD CEMENT PASTE. 
 
 A. (a) To manufacture a standard paste 1 kilogram of cement 
 will be used, which will be spread on a marble table in the shape 
 of a crown, in the center of which there will be poured out at once 
 the volume of water necessary to satisfy the following conditions. * 
 
 According to the nature of the tests, fresh or sea water may be 
 employed. 
 
 The mixture should be worked briskly with a trowel for five 
 minutes, counting from the moment when the water was poured 
 out. 
 
 (b) With one portion of the paste obtained there will be filled 
 immediately a flat-bottomed metallic box, truncated in form, hav- 
 ing a diameter of 8 centimeters at the lower base, 9 centimeters at 
 the upper base, and 4 centimeters deep; the surface will be 
 smoothed off by scraping a trowel along the upper surface of the 
 mold, avoiding any heaping and any shaking. 
 
 (c) In the center of the mass thus formed there will be brought 
 down normal to the surface of the paste, carefully and without 
 
 * That volume ought to be determined by means of successive trials. 
 
20 TESTS OF CEMENTS. 
 
 allowing it to acquire any velocity, a cylindrical sound 1 centi- 
 meter in diameter and weighig 300 grams, made of polished metal, 
 clean and dry, and terminated by a section normal to the axis. 
 
 That apparatus, called a " consistency sound," should be so con- 
 structed as to permit the exact determination of the thickness of 
 the paste remaining between the bottom of the box and the lower 
 extremity of the sound. Two tests will never be made on the 
 paste contained in the same box. 
 
 (d) The paste will be considered standard whose consistency is 
 such that the distance remaining between the bottom of the box 
 and the end of the sound, at the moment when the latter ceases to 
 sink under the action of its own weight, is 6 millimeters. 
 
 B. For rapid-setting cements the quantity of cement used in ex- 
 periments will be reduced to 500 grams and the duration of gauging 
 to 1 minute. 
 
 2. STANDARD MORTARS. 
 
 A. (a) To manufacture standard mortars, use will be made of 
 the natural sand coming from the shore of Leucate (Aude), suitably 
 screened, which will be called standard sand. According to cir- 
 cumstances, either simple standard sand or compound standard 
 sand will be used. 
 
 (b) Simple standard sand will be formed of grains passing a 
 plate screen perforated by holes 1.5 millimeters in diameter and 
 retained on one with holes I millimeter in diameter. 
 
 (c) Compound standard sand will be formed by a mixture of 
 equal weights of the following sands : 
 
 No. 1, whose grains, passing a screen of 2 millimeters, are retained 
 by a screen of 1.5 millimeters. 
 
 No. 2, whose grains, passing a screen of 1.5 millimeters, are re- 
 tained by a screen of 1 millimeter. 
 
 No. 3, whose grains, passing a screen of 1 millimeter, are retained 
 by a screen of 0.5 millimeter. 
 
 B. (a) For tests other than those of rupture, standard plastic 
 mortar will be used, and for rupture tests standard dry mortar. 
 
 (b) Standard mortars will be mixed in the ratio of 1 part by 
 weight of cement to 3 parts by weight of sand and will be gauged, 
 according to the nature of the tests, with fresh or with sea water. 
 
 One kilogram of material (250 grams of cement and 750 grams 
 of sand) is mixed intimately when dry; it is then formed on a 
 marble table in the shape of a crown, in the center of which there 
 will be poured out at one time the quantity of water to be employed, 
 and the mixture will be worked briskly with a trowel for five 
 minutes. 
 
TESTS OF CEMENTS. 21 
 
 (c) For the manufacture of standard dry mortar, use will be 
 made of simple standard sand. The quantity of water employed 
 in gauging will be 45 grams, augmented by a sixth of that neces- 
 sary to bring a kilogram of cement to the state of a standard 
 paste. 
 
 (d) For the manufacture of standard plastic mortar, use will be 
 made of compound normal sand. The quantity of water used in 
 gauging will be such that the resulting mortar has a plastic con- 
 sistency. To be assured that this consistency is well realized, a 
 metallic box, designed for tests of consistency, will be filled with a 
 part of the mortar obtained see above 6, No. 1, A (6) and 
 the surface will be planed off and smoothed with a trowel ; the 
 consistency will be considered satisfactory if, after smoothing, the 
 mortar exudes a little under the effect of several blows of a trowel 
 struck on the sides of the box. 
 
 C. For rapid-setting cements the quantity of cement used in 
 the experiments will be reduced to 500 grams and the duration of 
 gauging to one minute. 
 
 D. It is recommended for tests of mortar, other than standard 
 mortars, that there should be employed, in preference to all others, 
 mixtures of 1 part by weight of cement to 2 parts of standard 
 sand (rich mortars) and 1 part of cement to 5 parts of standard 
 sand (poor mortars).* 
 
 The first of those mixtures is particularly useful for rapid-setting 
 cements, with a view to completing the information furnished by 
 the standard mortars mixed 1 to 3. 
 
 RECOMMENDATION. The section recommends that, after an inter- 
 national agreement, standard plastic mortars be employed for tests 
 of every nature, to the exclusion of standard dry mortars. 
 
 7. TESTS OF SET. 
 
 Under the reservations expressed below, the section recommends 
 the adoption of the following rules for the determination of set of 
 pastes and mortars, f 
 
 1. PASTES. 
 
 A. (a) In pastes, both the beginning and the end of set will be 
 determined. 
 
 * P being the weight of water necessary to bring a kilogram of cement to the 
 condition of standard paste, the weight of water to employ should be 45 grams 
 plus f of P in the case of a dry mortar mixed 1 to 2, and 45 grams plus of P in 
 the case of a dry mortar mixed 1 to 5, the sand employed being always simple 
 standard sand. 
 
 f It is remembered that for each one of the tests concerning pastes and mor- 
 tars the conclusions relating to the standard test (intended to characterize the 
 products) are stated, following those relating to the corresponding general test. 
 
22 TESTS OF CEMENTS. 
 
 (b) At the moment of gauging, the temperature of the cement, 
 the water, and the air should be comprised between 15 and 18 C. 
 Immediately after its mixture, the paste, with the precautions 
 indicated hereinbefore in 6 (1, A), will be introduced into a box 
 similar to that described in the same paragraph (1, A b). As soon 
 as it is filled the box will be immersed in a tank containing water, 
 and the temperature will be kept between 15 and 18 C. The box 
 will be taken out of the tank only for the time necessary for each 
 determination. 
 
 (c) For the tests there will be employed a metal needle, called the 
 Vicat needle, which weighs 300 grams and which is cylindrical, 
 smooth, clean, dry, and terminated by an end cut at right angles 
 to the axis and containing an area of 1 square millimeter (diameter, 
 1.13 millimeters). The beginning of set will be called the instant 
 when the needle, descending normally to the surface of the paste, 
 carefully and without being allowed to acquire any velocity, can 
 no longer penetrate entirely to the bottom of the box. The end 
 of set will be called the instant when the surface of the paste can 
 support the same needle without penetrating into it any appreci- 
 able distance. The corresponding durations will be counted from 
 the moment when the gauging water has been placed in contact 
 with the cement. 
 
 (d) In the case of determining set in air, operations will be con- 
 ducted as has just been indicated, with this difference, that the 
 box as soon as it is filled will be kept in air at a temperature 
 between 15 and 18 C. ; care will be taken to void thoroughly the 
 water as it rises to the surface of the paste and separates from it. 
 
 B. The standard test of set will be conducted on the standard 
 cement paste immersed, as has just been described (A b). 
 
 2. MORTARS. 
 
 A. (a) In mortars, the end of set only will be determined. 
 
 (b) At the moment of tempering or gauging, the temperature of 
 the cement, sand, and water and the surrounding air should be 
 between 15 and 18 C. Immediately after its manufacture the 
 mortar will be introduced into the box serving for tests of con- 
 sistency ( 6, 1, A b) and will be struck and smoothed. As soon 
 as it is filled the box will be kept between 15 and 18 C. The box 
 will be taken out of the tank only for the time necessary for each 
 determination. 
 
 (c) When it is desired to determine the set in air, operations will 
 be conducted as has just been described, with this difference, that 
 the box, as soon as it is filled, will be kept in air at a temperature 
 between 15 and 18 C. 
 
TESTS OF CEMENTS. 23 
 
 (d) The end of set will be determined by the moment when the 
 surface of the mortar can support without deformation the pres- 
 sure of the thumb. The duration of set will be counted from the 
 moment when the gauging water has been placed in contact with 
 the mixture of sand and cement. 
 
 B. (a) The standard test of set will be conducted on the normal 
 plastic mortar immersed as above described (A b). 
 
 (b) The end of set will be determined by the moment from which 
 the consistency sound (par. c, 1, Ac) loaded with 5 kilograms 
 brought down normal to the surface of the water, carefully and 
 \vithout allowing it to acquire any velocity, no longer penetrates 
 any appreciable distance. 
 
 RECOMMENDATIONS. (a) The above rules (1 and 2) apply to 
 rapid-setting cements as well as to slow-setting cements. For the 
 first, at least, the use of the thermometer ought to furnish useful 
 information. The section recommends that studies on the subject 
 be continued. 
 
 (b) The section also recommends that the studies which have 
 been undertaken relative to the set of cement pastes and mortars 
 be continued. 
 
 8. RUPTURE TESTS BY TENSION. 
 
 The section recommends the adoption of the following rules for 
 determining the resistances of cement pastes and mortars to 
 tension : 
 
 A. (a) For rupture tests by tension, use will be made of bri- 
 quettes of the form of a figure 8, called standard briquettes, 
 having a section in the middle of 5 square centimeters and of the 
 type of fig. 1 (p. 24). 
 
 (b) Molds presenting within the form of the briquettes will be 
 placed on a plate of marble or of polished metal after having been, 
 as well as the plate, well cleaned and rubbed with greasy linen. 
 Those molds will be filled, six at one time, from the same gauging, 
 in the case of slow-setting cements and four at one time in the case 
 of rapid-setting cements, putting, at one time, in each mold enough 
 material to make it run over. It will be tapped with the finger, 
 that no voids be left, and several blows with the trowel will be 
 struck on the sides of the molds, to complete filling and to facili- 
 tate the escape of air bubbles. Then it will be smoothed off by 
 passing the blade of a straight knife almost horizontally over the 
 edges of the mold in such a manner as to take away all the 
 excess without exerting any pressure. Finally the surface will be 
 smoothed by passing over it a knife resting on its edge. If a 
 cement paste is being operated upon, it must not be taken out of 
 the molds until it has acquired sufficient consistency. 
 
TESTS OF CEMENTS. 
 
 (c) The briquettes will be taken out of the molds by sliding the 
 molds on the plate, uiiclamping them, and taking from them the 
 briquettes without raising them, at the end of twenty-four hours, 
 counting from the beginning of gauging, and before, if necessary, 
 
 in any case where set has certainly terminated. In all cases, dur- 
 ing this delay of twenty-four hours, the briquettes should be kept 
 in an atmosphere saturated with humidity, sheltered from air cur- 
 rents and the direct rays of the sun, at a temperature comprised, 
 as near as possible, between 15 and 18 C. The delay of twenty- 
 four hours will be reduced to that of one hour for rapid-setting 
 cement pastes and to three hours for mortars of the same cement. 
 
 (d) It is recommended to weigh the briquettes after taking them 
 out of the molds if one wishes to be assured of the regularity of 
 their manufacture. 
 
 (e) At the expiration of the delays fixed above in paragraph c, 
 the briquettes will be immersed in the medium chosen for their 
 storage. If the briquettes are immersed in fresh water, the depth 
 of water in a tank should not exceed 1 meter, and that water should 
 be renewed every week. If they are immersed in sea water, 
 renewal should take place every two days during the first week, 
 
TESTS OF CEMENTS. 
 
 25 
 
 and after that every week. During the first week, the volume 
 occupied by the water in the tank should be equal to four times, at 
 least, that of the briquettes. In every case the nature of the stor- 
 age water will be specified. If the briquettes are preserved in air, 
 its hygrometric state will be kept as near as possible to that of 
 saturation, and they will be placed under shelter from currents of 
 air and from the direct rays of the sun. The temperature of the 
 medium (air or water) will be maintained as near as possible 
 between 15 and 18 C. 
 
 (/) The apparatus for rupture will be arranged in such a way 
 that the effort of tension exerted on the briquettes can be contin- 
 uous, and increase at the rate of 5 kilograms per second. The 
 form and the method of attaching the clips should conform to the 
 following sketch, which reproduces the arrangement in actual use : 
 
 . Z 
 
26 TESTS OF CEMENTS. 
 
 (g) Breaking will be done at the end of 7 days, 28 days, 3 
 months, 6 months, 1 year, 2 years, etc., counting from the gauging. 
 It will be done also at the end of twenty-four hours for mortars of 
 rapid-setting cements, and at the end of from three to twenty-four 
 hours for cement pastes of that nature. 
 
 (h) Briquettes coming from the same gauging will be divided as 
 much as possible between the different series of six, which will be 
 broken at the periods of tests enumerated in the preceding para- 
 graph. The results obtained in each test will be rendered for each 
 one of the six briquettes ; their mean will be formulated, and any 
 anomalies will be indicated. The results will be expressed by 
 saying that "the resistance to tension measured by operating on 
 standard briquettes in the shape of a figure 8, 5 centimeters square 
 in cross section, is so many kilograms per square centimeter." 
 
 B. (a) Standard tests of rupture by tension will take place 011 
 standard cement paste and on standard dry mortar preserved in 
 fresh water. For those tests, the general rules under "A" and the 
 special rules hereinafter will be conformed to so far as concerns 
 the manufacture of briquettes. 
 
 (b) At the moment of mixing, the cement, the sand, the air, and 
 the water will be at a temperature comprised between 15 and 18 C. 
 The standard dry mortar will be rammed into a mold with a spat- 
 ula of iron about 35 centimeters long, including the handle, pre- 
 senting a striking surface of 25 square centimeters, and weighing 
 250 grams. Quick little blows will be given, first on the circum- 
 ference of the briquettes, then on the center; finally more ener- 
 getic blows will be given, following always the same method of 
 procedure and continuing the ramming until the mass commences 
 to have a little elasticity and the water oozes to the surface. Cut- 
 ting off and smoothing will then be done, as has been explained 
 before (A 6). 
 
 9. RUPTURE TESTS BY COMPRESSION. 
 
 Under the reservation expressed below, the section recommends 
 the adoption of the following rules to determine the resistance of 
 cement pastes and mortars to compression. 
 
 A. (a) For rupture tests by compression there will be taken as 
 briquettes the half briquettes separated by tension. Each half 
 briquette will be crushed separately, but the total will be taken of 
 the results furnished by the two half briquettes. When half bri- 
 quettes are lacking, use can be made of cylindrical test pieces 45 
 millimeters in diameter and 22 millimeters high, made and pre- 
 served similar to the briquettes destined for tests of rupture by 
 tension ( 8, A). 
 
TESTS OF CEMENTS. 27 
 
 (6) Briquettes which are rough or present conspicuous protuber- 
 ances will be planed by a light rubbing by hand on a slab of grit- 
 stone. 
 
 (c) The breaking apparatus will be placed so that the effort of 
 compression may increase in a continuous manner, leading to the 
 crushing of a half briquette at the end of one or two minutes. 
 
 (d) Tests will be made at the epochs fixed for those of rupture 
 by tension, and will take place, like them, on a series of six briquettes. 
 
 (e) The results will be rendered for each one of the six double 
 briquettes (made up of two twin briquettes) submitted to the test. 
 At the same time their mean will be formulated and the anomalies 
 will be set forth. The results will be expressed by saying that 
 " the resistance to crushing measured by operating on half-standard 
 briquettes, shaped like a figure 8, is so many kilograms per square 
 centimeter."* 
 
 B. The standard tests of rupture by compression will take place 
 on the standard briquette pastes and standard dry mortars which 
 have served in the standard tests for rupture by tension. When 
 half briquettes are lacking, cylindrical test pieces 45 millimeters in 
 diameter and 22 millimeters high, made and kept in the manner 
 indicated for those tests, may be employed ( 8 B). 
 
 C. For tests having for their object a comparison of mortars 
 with other materials, it is recommended provisionally to employ a 
 cube of 50 square centimeters face area, placed on its side, f 
 
 In a general way the rules adopted for other materials will be 
 adhered to in these tests. 
 
 RECOMMENDATION. The section recommends that researches 
 should be continued with a view to studying the advantages which 
 might arise from the use of cylinders of small dimensions for com- 
 pression tests, as well as the substitution of punching for crushing. 
 
 10. RUPTURE TESTS BY BENDING. 
 
 The section proposes to recommend rupture tests by flexure, 
 which it would be of interest to introduce into current laboratory 
 practice. The rules to adopt will be the following : 
 
 A. (a) For rupture tests by flexure, test pieces in the form of 
 prisms 12 centimeters long, with a square section of 2 centimeters 
 on a side, will be employed. 
 
 (b) The preceding directions as to manufacture, taking out of 
 molds, weighing and preservation of tension-test briquettes, as well 
 as the periods of test, are applicable also to flexure tests. 
 
 * The surface of a briquette (by which it is necessary to divide the total load 
 of rupture) is 31 square centimeters. 
 
 f That is, placed in such a way that the pressure is exerted normal to one of 
 the faces, which has been in contact with the sides of the mold. 
 
28 TESTS OF CEMENTS. 
 
 (c) The test piece to be broken will rest on one of its lateral faces 
 which has been in contact with the mold, on two knife-edges dis- 
 tant from each other 10 centimeters; the load will be applied in the 
 middle by the aid of a slightly-rounded knife-edge. The apparatus 
 of rupture will be placed in such a way that the force exerted on 
 the test piece can increase in a continuous manner at the rate of 1 
 kilogram per second. 
 
 (d) Results, as in the case of rupture tests by tension and by 
 compression, will be rendered for each one of the six test pieces sub- 
 mitted to the tests ; at the same time their mean will be formulated 
 and the anomalies presented indicated. The results will be expressed 
 by saying that ' ' the load of rupture by flexure is so many kilo- 
 grams for a prismatic specimen with square cross section of 2 centi- 
 meters on a side, placed on two supports, distant from each other 
 10 centimeters." 
 
 B. Standard tests of rupture by flexure will be conducted on 
 pastes and mortars made and preserved as has been indicated for 
 standard tests by tension ( 8, B). 
 
 11. DEFORMATION TESTS. 
 
 Under the reservation expressed below concerning hot tests, the 
 committee recommends the adoption of the following rules : 
 
 A. Tests intended to show deformations caused by the presence 
 of expansive materials will be conducted on cement pastes, either 
 cold or hot. 
 
 B. COLD TESTS. (a) For those tests the paste will be spread 
 out on a glass plate in such a manner as to form a cake about 10 
 centimeters in diameter and 2 centimeters thick, thinning out 
 toward the edges. Immediately after their manufacture, cakes 
 intended for tests in water will be immersed in the same condi- 
 tions as the test pieces intended for rupture tests ( 8, Ae). 
 
 Cakes intended for air tests will also be exposed under the con- 
 ditions indicated for such test pieces ( 8, Ae). 
 
 There will be noted the condition of the cakes at the ends of 
 periods of time assumed for rupture tests (7 days, 28 days, 3 months, 
 6 months, 1 year, 2 years, etc.). 
 
 (b) If it is intended to measure the swelling which is caused in 
 pastes of cement by the effect of prolonged immersion in cold water, 
 there can be used rods 80 centimeters long, with a square section 
 of 12 millimeters on a side, which will be placed vertically in glass 
 tubes 25 millimeters in diameter, filled with water. The elonga- 
 tion will be determined by the displacement on a scale of a needle 
 actuated by a stem which has been sealed to the upper extremity 
 of the prism. 
 
TESTS OF CEMENTS. 29 
 
 C. HOT TESTS. (a) For those tests there will be employed cyl- 
 indrical test pieces 3 centimeters in diameter and 3 centimeters 
 high, made in metal molds 0.5 millimeter thick. Each mold will 
 be slit in the direction of its axis and will carry, soldered to each 
 side of that slit, two needles 15 centimeters long; the increase in 
 the departure of the extremities of those two needles will measure 
 the swelling. 
 
 (b) The molds, as soon as they are filled, will be immersed in 
 cold water. As soon as the set is finished, and after a delay 
 which will not exceed twenty-four hours beyond set, the tempera- 
 ture of the water will be increased progressively to 100 C., the 
 time being comprised between one-fourth and one-half hour. A 
 temperature of 100 will be maintained for six consecutive hours, 
 and it will be allowed afterwards to cool, in order that the final 
 measurements may be made. 
 
 (c) NOTE. This method of hot tests is not applicable to rapid- 
 setting cements. 
 
 D. The standard tests of deformation will be made on a stand- 
 ard paste of cement. 
 
 RECOMMENDATION. The section recommends that prolonged 
 experiments be made through several years and conducted on a 
 great number of cement samples, notably on rapid-setting cements, 
 with a view to furnishing more complete information than that 
 now at hand concerning the comparative deformation in tests exe- 
 cuted hot or cold. 
 
 12. YIELD TESTS. 
 
 The section recommends the adoption of the following rules : 
 
 A. (a) The yield of a cement paste is the volume of the paste 
 obtained by gauging at normal consistency 1 kilogram of cement. 
 The yield in mortar of a cement is the volume of mortar obtained 
 by gauging at a plastic consistency 1 kilogram of sand and cement, 
 mixed in the proportions corresponding to that mortar. 
 
 (6) The yield will be determined by noting the volume occupied 
 in a graduated, cylindrical glass test tube, about 6 centimeters in 
 diameter, by the paste of the mortar which is introduced into it 
 immediately after gauging, with the precautions necessary to avoid 
 as far as possible the imprisonment therein of air bubbles. 
 
 (c) If necesary the yield can be determined with more precision 
 by molding the paste or the mortar in a block of any form what 
 ever and determining, after hardening, the difference in weight in 
 air and in water of such a block first dried from ooze water. 
 
 B. The standard test of yield will be conducted on the standard 
 cement paste and on the standard plastic mortar. 
 
30 TESTS OF CEMENTS. 
 
 13. POROSITY TESTS. 
 
 While remarking that the porosity of mortars and pastes, although 
 capable of being defined with precision, can not be measured with 
 rigorous exactitude, the section recommends, concerning those 
 measurements, the adoption of the following rules : 
 
 A. The porosity of a paste or of a mortar has for its measure 
 the ratio of the volume of voids which that paste or that mortar 
 presents to the apparent total volume, such voids including the 
 volume occupied by the water of absorption and by the hygrometric 
 water, but excluding the water of crystallization which evidently 
 forms a part of the solid. If we call V the apparent total volume 
 and v the solid volume, the porosity is determined then by the 
 formula : 
 
 V-v 
 Porosity = ~~y' 
 
 B. (a) To determine the porosity, operations will be conducted 
 on specimens having, as far as possible, an apparent volume com- 
 prised between y 3 ^ and T \ liter. 
 
 (6) The solid volume (v) will be obtained by taking the differ- 
 ence (P p) in the weight of the dry test piece, weighed in air (P), 
 and the weight of the test piece saturated with water and weighed 
 in water (p). 
 
 To obtain complete saturation, the test piece will be kept for a 
 quarter of an hour in air rarefied to a pressure not exceeding 25 
 millimeters of mercury, and the water will be made to arrive on 
 the test piece until its complete immersion under the same degree 
 of rarefaction. The test piece once covered with water, the atmos- 
 pheric pressure will be reestablished, and twenty-four hours will 
 elapse before obtaining the weight given as p. 
 
 When convenient means for rarefying the air are lacking, complete 
 saturation will be produced by the action of boiling water, when the 
 mortars can support such action without bad results. For that 
 purpose the test piece will be allowed to remain in water for forty- 
 eight hours ; at the end of that time it will be completely immersed 
 in cold water, which will be brought to the boiling point and main- 
 tained in that condition for two hours. Then it will be allowed 
 to cool without taking out the test piece, and at the end of twenty- 
 four hours will be obtained the weight which is given as p. To 
 dry the test piece it will be kept, until it no longer loses weight, 
 in a chamber heated to between 40 and 50 C. The final 
 measured weight will be P. For this operation care will be taken 
 that there does not enter into the oven any carbonic acid coming 
 from the products of combustion in the heating apparatus. For 
 certain products, drying effected under those conditions can not 
 
TESTS OF CEMENTS. 31 
 
 make all of the hygrometric water disappear, or may, on the con- 
 trary, take away a little of the water of crystallization, which per- 
 mits a slight uncertainty in the values found for porosity. 
 
 (c) The apparent volume of the test piece (V) can be obtained 
 by direct measurement, if it presents a geometrical form. If not, 
 the volume will be measured by taking the difference between the 
 weight of the test piece in water and in air, its condition of satura- 
 tion remaining the same. To insure the constancy of that state of 
 saturation, the test piece will be covered with a thin layer of melted 
 grease, which will be placed on with a brush and spread out with 
 the fingers. Care will be taken to weigh it in water before it is 
 weighed in air. 
 
 C. (a) The standard test of porosity will take place on the stand- 
 ard plastic mortar twenty-eight days old, preserved in water. 
 
 (b) For tests made on mortars of different age and composition, 
 it is recommended to employ, preferably, plastic mortars mixed 1 
 to 2 and 1 to 5, and aged 7 days, 28 days, 6 months, and 1 year. 
 
 (c) In all cases there should be indicated the composition, the 
 age, and the methods of preservation of the mortars submitted to 
 the tests. 
 
 14. PERMEABILITY TESTS. 
 
 The section has not been able, on account of insufficient informa- 
 tion, to formulate any proposition for the measure of permeability 
 of pastes and mortars so far as gases are concerned. For per- 
 meability tests in the case of water, the adoption of the following 
 rules is recommended : 
 
 A. (a) The permeability of pastes and mortars will be expressed 
 by the number of liters of water flowing per hour through a cubic 
 block 50 centimeters square on a face, under the following condi- 
 tions : 
 
 (b) The water intended for filtration will be led by a glass tube 
 35 millimeters in diameter and 11 centimeters high, sealed verti- 
 cally by the aid of pure cement to the upper face of the block 
 placed on its side,* which has been previously scraped to remove 
 all foreign matter. The tube, closed at its upper end by a rubber 
 stopper, will be put in communication with a reservoir elevated to 
 a level corresponding to the desired pressure of water. 
 
 There will be adopted for this pressure, according to the per- 
 meability of the mortars, heights of 10 centimeters, 1 meter, or 10 
 meters, f 
 
 * See footnote to 9. 
 
 f Where it is desirable to adopt different heights, multiples of \ meter will be 
 chosen preferably. 
 
32 TESTS OF CEMENTS. 
 
 (c) Before being submitted to the test the block will be immersed 
 in a tank during forty-eight hours, with precautions necessary to 
 arrive at as complete saturation as possible. Once under test, the 
 block will be kept immersed up to its full height. 
 
 (d) The volume flowing per hour will be determined after 24 
 hours, 7 days, 28 days, 3 months, etc.* 
 
 (e) The determination will take place on three similar blocks ; 
 mean results will be given corresponding only to the two blocks 
 that are most concordant. In rendering the results for permea- 
 bility at different epochs (24 hours, 7 days, 28 days, 3 months, etc.) 
 there must be given the pressure ( T V meter, 1 meter, or 10 meters) 
 under which tests were made. 
 
 B. (a) The standard test of permeability will take place on the 
 standard plastic mortar twenty-eight days old, preserved in water. 
 
 (b) For tests made on mortars of different age and composition 
 it is recommended to employ preferably plastic mortars mixed 1 to 
 2 and 1 to 5, aged 7 days, 28 days, 3 months, etc. 
 
 RECOMMENDATION. The section recommends that studies be 
 instituted with a view to determining methods of testing for per- 
 meability of mortars so far as gases are concerned. 
 
 15. TESTS OF DECOMPOSITION BY SEA WATER. 
 
 Although the conditions in which laboratory tests for decom- 
 position of pastes or mortars by sea water are made generally ex- 
 aggerate effects, such tests can give useful indications and are 
 recommended. The section proposes to adopt, concerning those 
 tests, the following rules : 
 
 A. (a) The tests will be made by immersion and by filtration. 
 
 (b) There will be employed for immersion standard briquettes 
 shaped like a figure 8, kept twenty-four hours after their manu- 
 facture in a tank containing sea water, which will be renewed 
 every two days for the first week and afterwards every week. 
 
 During the first week the volume of water should be equal to at 
 least four times that of the briquettes. 
 
 (c) There will be employed for filtration test pieces in the form 
 of cubical blocks similar to those designed for permeability tests, 
 arranged as has been indicated for such tests ( 14 A). The height 
 of pressure will be 10 centimeters, 1 meter, or 10 meters, according 
 to the permeability of the specimens submitted to the tests. Oper- 
 ations will be conducted on two series of test pieces ; those of the 
 first series will be kept in air, those of the second will be kept 
 immersed up to their full height in sea water. 
 
 *At the beginning of the tests the determinations will be multiplied if neces- 
 sary. 
 
TESTS OF CEMENTS. 33 
 
 (d) When natural sea water, is lacking use can be made of artifi- 
 cial sea water having the following composition : 
 
 Grams. 
 
 Sodium chloride (NaCl) 30 
 
 Crystallized sulphate of magnesium (MgOSO 3 , 7 HO) . . _ 5 
 
 Crystallized chloride of magnesium (MgCl, 6 HO) 6 
 
 Hydrated sulphate of lime (CaOSO 3 , 2 HO) 1. 5 
 
 Bicarbonate of potash (KOHO, 2 CO*) 0. 2 
 
 Distilled water, rain water, or boiled water 1,000 
 
 (e) Series of standard 8-shaped briquettes and cubical blocks 
 will be preserved in fresh water to serve for comparative tests. 
 
 (/) The results of the tests will be expressed by giving the fol- 
 lowing comparative information for the two series of briquettes : 
 
 (1) Modification of the appearance of the test pieces. 
 
 (2) Resistance to tension and to compression for the briquettes 
 immersed, and to compression for the blocks submitted to filtra- 
 tion. 
 
 (3) Chemical composition. 
 
 The tests will be made, according to circumstances, after one or 
 after several of the periods fixed for resistance to rupture (28 days, 
 3 months, 6 months, 1 year, etc.). 
 
 B. (a) The standard test of decomposition by sea water will take 
 place on the standard plastic mortar; for filtration tests, operations 
 will be conducted on specimens twenty-eight days old and kept in 
 sea water. 
 
 (&) For tests made on mortars of different age and composition, 
 it is recommended to use preferably mortars mixed 1 to 2 and 1 to 
 5, aged 7 days, 28 days, 3 months, etc. 
 
 (c) In all cases there will be indicated the composition, age, and 
 method of preservation of the mortars submitted to the tests. 
 
 16. TESTS OF ADHESION. 
 
 The section proposes to recommend for adhesion tests the adop- 
 tion of the following rules, stating that some among them are 
 based upon a comparatively restricted number of experiments 
 and can not be considered as definite : 
 
 A. To compare the adhesive strength of cements there will be 
 submitted to tests of rupture by tension specimens in the form of a 
 double T made by the aid of a special mold which is designed as 
 shown in fig. 3, each one of the two materials whose adhesion is 
 studied constituting one of the halves of each test piece. 
 
 For those tests, the following rules will be adhered to : 
 
 B. STANDARD TESTS FOR COMPARING THE ADHESION OF DIF- 
 FERENT CEMENTS TO THE SAME MATERIAL. (a) There will be 
 prepared standard adhesion blocks of mortars composed by weight 
 
 15671 3 
 
TESTS OF CEMENTS. 
 
 * 
 
TESTS OF CEMENTS. 35 
 
 of 1 part Portland cement, passing through a screen composed of 
 900 meshes, and 2 parts of standard sand No. 3. ( 6, 2, A c.) 
 
 The mortar will be gauged with 9 per cent of water and firmly 
 compressed into a mold whose bottom is provided with a movable 
 metallic plate. The adhesion blocks will be immersed in fresh 
 water at the end of twenty-four hours. When it is desired to use 
 them they will be dried. Then the adhesive surface will be passed 
 over by emery paper. 
 
 (6) There will be employed for this test the standard plastic 
 mortar which will be introduced by a simple beating with a trowel 
 into a mold placed in such a way that the standard adhesive block 
 will form the bottom. The taking of the test piece out of the 
 mold formed by the standard block attached to the mortar to be 
 tested, will be done as soon as set has completely taken place. 
 
 (c) The rules for tension tests should be followed as to the num- 
 ber and preservation of the test pieces, the periods of test, the 
 mode of rupture, and the expression of the results. 
 
 C. STANDARD TESTS FOR COMPARING THE ADHESION OF THE 
 SAME CEMENT TO DIFFERENT MATERIALS. (a) For those tests 
 the above directions will be followed, with this exception, that the 
 normal adhesion blocks will be replaced by blocks made of the 
 different materials to be tested. If the material can be molded 
 the adhesion block will be made in a mold like a standard block. 
 If the material is nonplastic there will be prepared a block several 
 millimeters in thickness, having a well-smoothed face, which will 
 be placed at the bottom of the mold, and the test block will be 
 completed by filling up the mold with neat cement. 
 
 (6) Where standard plastic mortar is not used, the composition 
 of the mortar employed will be stated. 
 
THE INFLUENCE OF SEA WATER 
 
 ON 
 
 HYDRAULIC MORTARS. 
 
 (37) 
 
THE INFLUENCE OF SEA WATER ON HYDRAULIC BINDING MEDIA.* 
 BY DR. WILLIAM MICHAELIS. 
 
 By allowing saturated limewater to act on the hydrates of silica, 
 ferric oxide, and alumina, there can be obtained the following com- 
 pounds of lime, representing the highest grades of saturation : 
 
 2 SiO f , 3 CaO + x H 2 O ; 
 2 Fe 2 O 3 , 4 CaO + y H 2 O; 
 2 A1,O 3 , 5 CaO + z H,O. 
 
 The values of x, y, and z have not been determined with certainty ; 
 they are assumed by me at 6, 7, and 8, respectively, the minimum 
 amount of water being 1 equivalent of it to 1 equivalent of CaO. 
 
 Until it has been demonstrated with certainty that there is a 
 difference between the compounds formed during the action of 
 water on calcareous hydraulic binding media and those formed dur- 
 ing the process of hydraulic set, it may be assumed that during 
 set the above enumerated compounds are actually formed, and 
 that the residual lime is eliminated as hydrate of lime, a process 
 evidently taking place in Portland cement, in which, when in its 
 hardened state, numerous crystals of hydrate of lime are found 
 interspersed. 
 
 As is well known, H. Le Chatelier assumes the following crys- 
 tallizing compounds, 2 CaOSiO 2 + 5 H 2 O, as the hydrosilicate of 
 lime, and A1 2 O 3 , 3 CaO + 12 H 2 O as the hydroaluminate of lime. 
 
 It is of no importance for the present discussion whether hydro- 
 silicate of lime is formed in crystals during hydraulic set, or 
 whether, on account of its absolute insolubility, it can not crys- 
 tallize, but is a colloid, f 
 
 It is well known that the compound A1 2 O 3 , 3 CaO expands con- 
 siderably in absorbing water. I do not, therefore, consider this 
 compound to be a predominating one in normal Portland cements. 
 
 * An abstract of a preliminary article by Dr. Michaelis on the same subject 
 was published in Vol. CVII, Minutes Proceedings Institution Civil Engineers, 
 1891. The present article appeared in Verhandlungen des Vereines zur Befdr- 
 derung des Gewerbfleisses, 1896. 
 
 Since the foregoing translation was sent to the printer, a copy of a translation 
 into English, printed at Edinboro, Scotland, has been received. 
 
 f Insolubility in this case means that in the presence of calcareous earth silicic 
 acid is entirely insoluble in water; hydrosilicate of lime, therefore, may be 
 decomposed by water, but is never dissolved ; the lime only passing into solution. 
 
 (39) 
 
40 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 No detailed information is given by Le Cnatelier concerning 
 hydroferrate of lime, a compound, by the way, which is exceed- 
 ingly unstable and readily decomposed; hence the existence of 
 none but the first-mentioned compound has been thoroughly 
 demonstrated. 
 
 In accordance with E. Candlot I have found the following for- 
 mula for the aluminate and sulphate of lime compound : 2 (A1 2 O 3 , 
 3 CaO) + 5 (CaOSO 3 ) + 80 H 2 O (dried over sulphuric acid) . Candlot 
 gives 120 H 2 O for the air-dried crystals. But the combination, 
 A1 2 O 3 , 3 CaO + 3 CaOS0 3 + 30 H 2 O, has also been observed by me.* 
 
 Only the second compound will be considered here ; that is, the 
 one crystallizing with 30 equivalents of water, which is certainly 
 less than is really the case. 
 
 Each part by weight of aluminate, which appears in hydraulic 
 mortars in the form of hydroaluminate of lime, is able to form 12 
 parts by weight of this double salt. 
 
 The hydroferrate of lime behaves similarly. I have found its 
 formula to be Fe 2 O.,, 3 CaO + 2 CaO, SO 3 + x H a O, but I have not 
 yet succeeded in obtaining it in crystallized form. Being yet of 
 a doubtful character, this compound will not be considered here 
 at all ; this much may, however, be assumed, that it acts similarly 
 to the above-mentioned double salt. 
 
 One part by weight of A1 2 O 3 will form 3.062 parts by weight of 
 2 A1 2 3 , 5 CaO + 8 H 2 O. 
 
 One part by weight of A1 2 O 3 will form 3.7 parts by weight of 
 ALA, 3 CaO + 6 H 2 O. 
 
 One part by weight of A1 2 O 3 will form 4.735 parts by weight of 
 A1 2 O 3 , 3 CaO + 12 H 2 O. 
 
 One part by weight of A1 2 O 3 will form 11.856 parts by weight of 
 sulphate and aluminate of lime. 
 
 One part by weight of A1 2 O 3 , 3 CaO + 6 H 2 O will form 3.2175 parts 
 by weight of sulphate and aluminate of lime, or 4.4636 parts by 
 weight of Candlot's double salt, with 120 H 2 O. 
 
 One part by weight of CaO, SO 3 will form nearly 3 parts by weight 
 of sulphate and aluminate of lime. 
 
 One part by weight of CaO, H 2 O will form nearly 2.3243 parts by 
 weight of CaO, SO 3 + 2H 2 O. 
 
 The true Roman cements, containing 1 part by weight of sili- 
 cate silicic acid, alumina, ferric oxide (manganous oxide) to 1.1 
 or 1.2 parts by weight of lime, are the best hydraulic mortars 
 
 *The air-dried compound, 2 A1 2 O 3) 3 CaO + 5 CaOSO 3 + 80 H 2 O, when dried 
 over sulphuric acid, has rendered 22 H 2 O [perhaps 2 (A1 2 O 3 , 3 CaO + 6 H 2 O) + 
 5 (CaO, SO 3 + 2 H a O)] ; dried at 100 C., it rendered 16 H 2 O, and heated to a dark 
 red, it became anhydrous. The anhydrous double salt is soluble in 2,214 parts 
 by weight of water at a temperature of 18 C. 
 
INFLUENCE OF SEA WATER OX HYDRAULIC MORTARS. 41 
 
 from a chemical point of view, because on hardening they form 
 the most stable compounds, without leaving any iionsaturated 
 residues. 
 
 For instance, a Roman cement of the following composition : 
 
 Silicic acid = 24.00 per cent or, in equivalents, 0.400; 
 
 Alumina = 10.28 per cent or, in equivalents, 0.200; 
 
 Ferric oxide = 4.80 per cent or, in equivalents, 0.030; 
 
 Lime = 49.00 per cent or, in equivalents, 0.875; 
 
 Sulphuric acid = 3.20 per cent or, in equivalents, 0.040; 
 
 Magnesia! = 5 . 00 percent; 
 
 Alkali 
 
 Residue = 3. 72 per cent; 
 
 100.00 per cent; 
 
 will require 0.4 X f+ 0.2 X f + 0.03 X 2 -f- 0.04 = 1.2 equivalents or 
 67.2 parts by weight of lime, when the calcareous compounds 
 mentioned at the opening of this article are considered. As the 
 cement contains only 0.875 equivalents of lime, the formation of 
 compounds as rich in lime as these is impossible, and compounds 
 will be formed that contain less lime and are therefore more sta- 
 ble: SiO 2 , CaO, 2 A1 2 O 3 , 3 CaO; 2 Fe 2 O 3 , 3 CaO. For the forma- 
 tion of these, only 43.96 parts by weight of CaO are required; 
 the remaining 5.04 parts of lime will then enter into compounds 
 richer in lime, and no free lime will remain after set. 
 
 A hydraulic mortar composed as above serves its purpose in the 
 best possible way. Leaving out of consideration for the present 
 the double compound of aluminate and sulphate of lime A1 2 O 3 , 
 3 CaO + 3 CaOSO 3 + 30 H 2 O, it would after set render the follow- 
 ing stable compounds : 
 
 Hydrosilicate of lime with from 1 to 1. 5 parts of CaO to 1 of SiO 2 ; 
 
 Hydroaluminate of lime with 3 parts of CaO to 2 of A1 2 O 3 ; 
 
 Hydroferrate of lime with 3 CaO to 2 of Fe 2 O 8 . 
 
 Only gypsum, hydrate of magnesia, and hydrate of caustic 
 potash will form an unsaturated residuum, the salts generated by 
 them with sulphuric acid being however readily soluble and not 
 injurious. 
 
 But as Roman cements are burnt at red heat, or at moderate red 
 heat only, during which operation they do not condense, they must 
 be pronounced from a physical point of view to be of a porous 
 nature ; the compounds forming during absorption of water will, 
 therefore, be contained in them in a much swollen state; hence 
 the mortars produced with them will shrink considerably during 
 air drying, through loss of loosely-bound water, nearly all water 
 contained in these hydrates, over and above the quantity corre- 
 sponding to the hydrate of lime, being such loosely-bound water. 
 
42 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 Physically the hydraulic limes, the best representative of which 
 is lime of Theil, are quite close to the Roman cements. Their 
 density, at least that of the so-called light ones, is generally less 
 than that of the Roman cements. 
 
 Freshly-calcined lime of Theil contains 
 
 Silicic acid = 22.80 per cent or, in equivalents, 0.3800; 
 Alumina = 2.57 per cent or, in equivalents, 0.0250; 
 
 Ferric oxide = 0.88 per cent or, in equivalents, 0.0055; 
 Sulphuric acid = 0.64 per cent or, in equivalents, 0.0080; 
 Lime = 68.60 per cent or, in equivalents, 1.225; 
 
 Magnesia = 1.60 per cent. 
 
 The most highly calcareous compounds forming during set will 
 use 0.38 x|+ 0.025X| + 0.0055X2 + 0.008 = 0.651 5 equivalents, or 
 36.48 parts by weight of CaO; consequently there remains an 
 unsaturated residue of 32.12 parts by weight of CaO aside from 
 1.6 parts by weight of magnesia. 
 
 From a physical point of view Portland cement is much superior 
 to the hydraulic limes, because it acquires great density through 
 vitrification at white heat. During set the pores are filled more 
 completely because, the particles being closer together, there is in 
 the same space much more swelling substance. The mean pro- 
 portion of mass in equal spaces is for Portland cement and for 
 Roman cement about as 5:3, and for Portland cement and hy- 
 draulic limes from 5 : 2.5 to 5 : 2; vitrified Portland cement, there- 
 fore, has a much greater volume weight, and consequently its 
 mortar attains a much higher degree of strength and density, or 
 rather of condensation, since Roman cements and hydraulic limes 
 may likewise present a perfectly close grain. 
 
 Chemically, however, Portland cements are inferior, because, 
 like lime of Theil, they leave during hydraulic set a considerable 
 residue of lime, striving for saturation. The following Portland 
 cement, for instance, approaches the lower limit in its percent- 
 age of lime : 
 
 Silicic acid 22.50 per cent or, in equivalents, 0.3750; 
 Alumina = 8.99 per cent or, in equivalents, 0.0875; 
 
 Ferric oxide = 4.00 per cent or, in equivalents, 0.0250; 
 Sulphuric acid = 1.00 per cent or, in equivalents, 0.0125; 
 Lime = 61.04 per cent or, in equivalents, 1.0900; 
 
 Magnesia ) 
 
 A11 e v = 2.47 per cent. 
 
 Alkali 
 
 100.00 per cent. 
 
 The most highly calcareous compounds in this case will require 
 0.84375 equivalents, or 47.25 parts by weight of lime; hence there 
 
INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 48 
 
 will remain at least 13.79 parts by weight of unsaturated lime, 
 aside from magnesia and alkali. 
 
 The following Portland cement approaches the upper limit in 
 its percentage of lime : 
 
 Silicic acid = 20.778 per cent or, in equivalents, 0.3463; 
 Alumina = 5.819 per cent or, in equivalents, 0.0566; 
 
 Ferric oide = 2.720 per cent or, in equivalents, 0.0170; 
 Sulphuric acid = 0.520 per cent or, in equivalents, 0.0065; 
 Lime = 68.379 per cent or, in equivalents, 1.2210; 
 
 Magnesia) = m 
 Alkali j 
 
 If we assume the most highly calcareous compounds as having 
 actually formed during set in this case, a residue of 29.1 parts by 
 weight of unsaturated lime will be left. 
 
 In a Portland cement of average composition (1 part by weight 
 of silicate to 2 parts by weight of lime) 25 per cent of lime, or 33 
 per cent of hydrate of lime, is thus separated. As for the portion 
 of the hydrate which does not become carbonate of lime, it is 
 plainly discernible in all Portland cement mortars in a crystal- 
 lized state. 
 
 It is readily understood that a compound in which such a con- 
 siderable amount of caustic lime becomes free can not be a chemi- 
 cally stable compound in the beginning. The free lime will react 
 and work until, in one way or another, it has entered into a satu- 
 rated compound. At first, this is done through absorption of car- 
 bonic acid from the surface, if the mortar is in contact with air 
 or water containing carbonic acid ; in sea water the lime will be 
 acted upon principally by the soluble compounds of sulphuric 
 acid. To begin with, the completely free lime will become car- 
 bonate of calcium and sulphate of calcium; after that the lime 
 contained in the exceedingly unstable ferric-acid compound will 
 undergo the same change ; the aluminate of lime is then attacked, 
 and finally the silicate. 
 
 The simple formation of sulphate of lime, with 2 equivalents of 
 water, implies a considerable increase of volume, and in itself is 
 sufficient to destroy the cohesion attained during the absorption 
 of water. But together with this there occurs the formation of 
 sulphate and aluminate of lime, which implies an immense increase 
 of volume and a consequent total destruction of all cohesion ; this 
 double compound crystallizes with at least 30, but probably with 
 60, equivalents of water, and in doing so cleaves grain by grain 
 of the strongest mortar with irresistible power, leaving merely an 
 incoherent slime, in which only such portions may contain a cer- 
 tain amount of cohesion as have been protected by the formation 
 of carbonate of calcium. 
 
44 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 On examining the Roman cements, the hydraulic limes, and the 
 Portland cements in regard to this formation of sulphate of lime, 
 and subsequently of aluminate and sulphate of lime, it is found 
 that in Roman cements all lime is bound, and has no tendency to 
 combine with sulphate of magnesium. The 5.44 parts by weight 
 of sulphate of calcium present in the mortar, it is true, are able 
 to form at least 16 parts by weight of the double salt, that is, an 
 increase of about 11 parts by weight of solid substance; but it is 
 almost certain that there is ample space for this increase of sub- 
 stance in the pores of the mortar. According to my observations, 
 and the experience of others, good Roman cements will resist the 
 action of sea water excellently. 
 
 In lime of Theil the free lime generally has been carbonated to 
 a great extent through hardening in air before immersion in sea 
 water, and thereby is withdrawn from the possibility of forming 
 gypsum. Assuming that no appreciable formation of carbonate 
 of calcium has taken place, then about 30 parts by weight of lime 
 would be available for the formation of gypsum, and out of 39.643 
 parts by weight of hydrate of lime, 92.144 parts by weight of gyp- 
 sum might be generated, which by itself would suffice to destroy 
 the mortar; but, in addition to this, the existing 5.97 parts by 
 weight of aluminate of lime, with a portion of the gypsum, would 
 form about 30 parts by weight of the double compound, making 
 an entire increase of about 60 parts by weight of solid substance, 
 viz, 30 parts by weight of the double salt plus 92 parts by weight 
 of hydrate of lime, 9.25 parts by weight of hydroaluminate of 
 lime, and 13 parts by weight of gypsum (employed in the forma- 
 tion of the double salt). 
 
 In this case the formation of gypsum plays the principal part ; 
 it alone renders an increase of substance of 52.5 parts by weight. 
 The formation of the double salt is comparatively insignificant, 
 hydraulic limes containing little alumina and consisting mainly of 
 lime and silica. The double salt, however, undoubtedly has a 
 much greater force of crystallization, consequently a greater force 
 of expansion. 
 
 In Portland cements with an average percentage of lime (64 
 per cent lime, 7.2 per cent alumina, and 0.8 per cent sulphuric 
 acid), there will separate, as we saw above, about 25 per cent of 
 CaO (33 parts by weight of CaH 2 O 2 ), which will combine with 
 sulphate of magnesium forming 74.5 parts by weight of gypsum, 
 thus causing an increase of substance of 41.5 parts by weight. 
 The existing 7.2 per cent of alumina had formed 26.64 parts by 
 weight of hydroaluminate of lime, and now will combine with the 
 existing gypsum, forming at least 85 parts by weight of double 
 salt. The total increase of substance in this case then amounts to 
 
INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 45 
 
 74.5 parts of gypsum plus 85 parts of double salt, minus the exist- 
 ing 33 parts of CaH 2 O 2 , the 26.64 parts of hydroaluminate of 
 lime, and 36 parts from the existing gypsum; that is, in all, an 
 increase of 64 parts by weight. This enormous increase of sub- 
 stance to about 125 parts by weight of water-hardened mortar 
 must infallibly cause its total destruction, unless the action of the 
 soluble sulphates upon each other is prevented by especially favor- 
 able circumstances. Such circumstances are partly of a chemical 
 and partly of a physical nature. 
 
 Under ordinary temperatures carbonate of lime can not be decom- 
 posed by sulphates ; hence the most extensive formation possible 
 of carbonate of calcium out of the excess of hydrate of lime will 
 be the best protection. On account of their slow setting, hydraulic 
 limes, when used for maritime constructions, are allowed to harden 
 in air for a long time before being submerged, and in this way are 
 enabled to absorb an abundance of carbonic acid ; hence, presum- 
 ably, the fact that lime of Theil, for instance, has behaved much 
 better than Portland cement, which latter in most cases succumbs 
 to the influence of sea water. 
 
 There are, therefore, good reasons for allowing concrete blocks 
 of Portland cement to harden in air for a long time before sub- 
 merging them, thus surrounding them with a protective shell in 
 which free lime has changed into carbonate of calcium. But only 
 a layer of insignificant thickness can thus be saturated with car- 
 bonic acid during a reasonable length of time, and if the sea water 
 at any subsequent period does penetrate to the core of the block, 
 then the above-described chemical action and destruction will 
 occur after all, and the outer shell of carbonate will be rent with 
 irresistible force and lifted off the rapidly-decaying interior. 
 
 Silicate of lime offers a fair resistance to the change into car- 
 bonate of lime, at least to the total change. Carbonic acid in this 
 case behaves precisely like water. All lime beyond 1 equivalent 
 is much more soluble in water and changes more readily into car- 
 bonate of lime; the less lime the more resistance the compound will 
 offer to the attack of carbonic acid. Even in small briquettes of 
 from 50 to 100 cubic centimeters, which are kept in a moist atmos- 
 phere of pure carbonic acid, it is very difficult to change all lime 
 into carbonate. 
 
 Aluminate of lime decomposes more- readily, and the ferrate of 
 lime is decomposed entirely, by the carbonic acid. 
 
 The following investigation of the influence of carbonic acid on 
 the strength and the resistance to sea water of Portland cement 
 mortar was made: 
 
 Ten tensile briquettes were formed of 1 part by weight of Stettin 
 Portland cement (Stern) and 3 parts by weight of Berlin standard 
 
46 
 
 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 sand, and the same number of briquettes of 1 part by weight of 
 the same cement and 5 parts by weight of the same sand. All 
 were allowed to harden in moist air (under glass bells) for twenty- 
 four hours, and after that in hermetically-sealed bottles in distilled 
 water for fifty-six days. Then half of the test pieces were kept 
 under a glass bell in moist carbonic acid for five weeks, at the end 
 of which period they were returned to those in the bottles and 
 kept under water for another four weeks. They were then 120 
 days old, and were tested for tension as well as compression, the 
 pieces broken by tension being held together by rubber bands for 
 the crushing test. The results (strains given in kilograms per 
 square centimeter) are as follows : 
 
 CEMENT MOKTAK, 1:3. 
 
 CEMENT MORTAR, 1:5. 
 
 a. PROTECTED AGAINST 
 C0 2 . 
 
 b. TREATED WITH C0 2 . 
 
 c. PROTECTED AGAINST 
 C0 2 . 
 
 d. TREATED WITH C0 2 . 
 
 Tensile 
 strength. 
 
 Crushing 
 strength. 
 
 Tensile 
 strength. 
 
 Crushing 
 strength. 
 
 Tensilo 
 strength. 
 
 Crushing 
 strength. 
 
 Tensile 
 Strength. 
 
 Crushing 
 strength. 
 
 26.5 
 
 301 
 
 27.5 
 
 336 
 
 13.5 
 
 115 
 
 15.5 
 
 139 
 
 25.5 
 
 323 
 
 30.0 
 
 364 
 
 13.5 
 
 115 
 
 20.0 
 
 150 
 
 28.0 
 
 330 
 
 28.5 
 
 374 
 
 14.0 
 
 96 
 
 16.5 
 
 130 
 
 28.0 
 
 357 
 
 29.0 
 
 322 
 
 14.5 
 
 105 
 
 15.5 
 
 138 
 
 26.0 
 
 237 
 
 27.5 
 
 296 
 
 14.5 
 
 132 
 
 16.5 
 
 133 
 
 26.9 
 
 309 
 
 28.5 
 
 338 
 
 14.0 
 
 112 
 
 16.8 
 
 138 
 
 How much of the lime contained in the cement has become car- 
 bonate of calcium in the case of b and d ? 
 
 What was the percentage of water in the mortars protected 
 against carbonic acid, and in those treated with it, after drying 
 over sulphuric acid ? 
 
 In regard to these questions the following was found for the 
 various mortars : 
 
 Per cent. 
 
 (Carbonic acid 0.565 
 
 t Water of crystallization 3 . 046 
 
 [Carbonic acid .. . 1.506 
 
 Mortar a. 1 
 
 Mortar b. 1:3 
 
 Mortar c. 1:5 
 
 iv/r ,7 1 
 
 [Water of crystallization 3. 150 
 
 [Carbonic acid 0.483 
 
 t Water of crystallization 2. 336 
 
 (Carbonic acid 1. 954 
 
 (Water of crystallization 1. 896 
 
 The 1:3 mortar contained 14.383 per cent and the 1:5 mortar 
 contained 10 per cent of lime, consequently under b only 13.3 per 
 cent and under d only 24.86 per cent of the entire amount of lime 
 had been carbonated. It is, therefore, evident that the absorption 
 of carbonic acid is a very slow process, even in such small solids as 
 
INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 47 
 
 tensile briquettes (70 cubic centimeters with, a height of 23 milli- 
 meters), placed in a moist atmosphere of that gas, and that it 
 requires a very long treatment to decompose the silicates and the 
 aluminates of lime. Hence, even a very long exposure of con- 
 crete blocks in open air can only carbonate them superficially, 
 and can not be looked upon as a sufficient protection against the 
 destructive action of sea water. 
 
 Only after long exposure in the powdered state to the action of 
 carbonic acid were the entire contents of lime carbonated ; this 
 shows that the action of carbonic acid may be likened to that of 
 fresh water; the less lime contained in the hydrosilicates and 
 aluminates the greater their power of resistance. 
 
 Some of the halves of the tensile briquettes were put into sea 
 water and a 2 per cent solution of sulphate of magnesia; those 
 which had been protected against absorption of carbonic acid 
 decayed very rapidly, but those which had been treated with car- 
 bonic acid were also considerably affected after only seven months ; 
 the more porous mortars of 1 : 5 visibly more, although they had 
 absorbed more carbonic acid. 
 
 Hydraulic mortars exposed to sea water should therefore consist 
 of the most stable compounds of hydrate of lime, with hydrate of 
 silica, alumina, and ferric oxide. 
 
 3 3 
 
 2 SiO, Z CaO and 2 A1 2 O 3 ^ CaO. 
 1 1 
 
 In fresh water the case is much more favorable ; here, the free 
 lime only can be either dissolved or carbonated. The more lime 
 is lixiviated by water, the less soluble the remaining lime (which 
 is bound to silicic acid or to alumina) becomes; this has been 
 demonstrated already by Le Chatelier, and my experiments on fully 
 hardened, neat Portland cement have verified it. The mortar may 
 have become more porous, but complete dissolving of all of the 
 lime will hardly ever occur. On a small scale Portland cement, 
 which has been protected against absorption of carbonic acid, may 
 be completely freed of its lime by digestion with boiled distilled 
 water, leaving only the hydrates of silica, alumina, and ferric oxide. 
 But pieces of only a few cubic centimeters volume require a treat- 
 ment of years with very large volumes of water, in order to effect 
 perfect decomposition. 
 
 In fresh water, therefore, the exact opposite of the above process 
 takes place; the lime partly exudes from the mortar and thus 
 leaves it, it is true, in a more porous state, but in a faultlessly- 
 made cement there will never develop strains which destroy the 
 cohesion of the mass. If Portland cement containing up to 70 per 
 cent of CaO hardens with water, then the binding force during 
 
48 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 the formation of hydrates is greater than the expansive force of 
 swelling if the separation of lime takes place uniformly through- 
 out the entire mass. Here the increase in volume of lime is incon- 
 siderable, as in caustic lime which has been mixed dry with sand and 
 then gauged with water. The increase in volume in these cases ap- 
 pears to take place in the pores, thus causing the mass to increase in 
 density. The matter is different when imperfect physical mixtures 
 are burned ; then the vitrified cement contains free lime (coarse 
 grains in the interior), also silicates, aluminates, and ferrates of too 
 highly calcareous a nature, of which those with 3 equivalents and 
 more of CaO to 1 equivalent of SiO 2 , A1 2 O 3 , and Fe 2 O 3 will show 
 during the absorption of water an increase of volume similar to that 
 of free lime. At any rate I have demonstrated by means of melted 
 Portland cement that there is a Portland cement entirely constant 
 in volume, consisting of 1 part by weight of silicate (silica, 
 alumina, ferric oxide (manganese)) to 2.4 parts by weight of lime, 
 or, according to French designation, with a factor of hydraulicity 
 of 0.416. 
 
 It is evident that closeness of grain in other words, impervious- 
 ness is a very important consideration. 
 
 Magnesia, which forms during the action of sea water on 
 hydraulic mortars, and which has erroneously been accused of 
 causing their destruction, so far from doing this is really a protec- 
 tion, since under the action of hydrate of lime on sulphate of mag- 
 nesia the insoluble hydrate of magnesia will form and aid in closing 
 pores, that is, in increasing imperviousness. 
 
 The injurious influence of sea water should no longer be attrib- 
 uted to the salts of magnesia, since such injury results only from 
 the sulphuric acid or from the soluble sulphates in the sea water. 
 Sulphate of magnesia, it is true, is the special sulphate active in 
 sea water, but sulphate of calcium, sulphate of any alkali in short, 
 every sulphate soluble in water has precisely the same character 
 of destructive influence, although not the same amount of energy. 
 
 Other mechanical means of protection consist in incrustations of 
 vegetation and animals or in layers of mud covering the surface 
 and preventing permanent action of sea water. 
 
 Not considering physical qualities, it has been demonstrated 
 through my investigations that the most highly calcareous 
 hydraulic binding media offer the least resistance to the action of 
 sea water, and that it is, therefore, a great mistake to add lime to 
 such mortars. The fact that mortars of Portland cement and of 
 hydraulic lime do not keep nearly as well as those of Roman 
 cements is proof sufficient that free lime is the principal cause of their 
 destruction by sea water. The higher the percentage of lime in 
 Portland cement, the more hydrate of lime remains free and striv- 
 ing for saturation and the greater is the increase of volume through 
 
INFLUENCE OF SEA WATER OX HYDRAULIC MORTARS. 49 
 
 formation of gypsum; hence the fact that modern Portland 
 cements of great strength are found less suitable for maritime con- 
 structions than the older, less calcareous ones. The cement quoted 
 as an example above, in which the percentage of lime is nearly the 
 practically possible maximum, would suffer a greater increase of 
 volume (70 parts by weight to 125) if all separated lime were to 
 change into sulphate of calcium and all hydroaluniinate of lime 
 into the double compound of aluminate and sulphate of lime with 
 only 30 equivalents of water, although in this case the amount of 
 the double salt formed would be less. 
 
 Hydraulic mortars with free caustic lime possess just as little 
 stability physically as they do chemically. Absorption of water, 
 even if it is only of a hygroscopic nature, causes swelling, and 
 consequently strong pressure strains as well as molecular changes. 
 To begin with, alkali becomes free and energetically absorbs water 
 and carbonic acid, the latter soon combining with the hydrate of 
 lime. As is well known, caustic lime can absorb carbonic acid in 
 the presence of free water ; dry carbonic acid does not act upon 
 dry hydrate of lime. But during each drying the higher hydrates 
 will lose water, tensile strains will develop, and shrinking and 
 cracking will occur. Lime surely does not form one hydrate only, 
 but hydrates may be assumed with 5 and 8 equivalents of water, 
 although it may be difficult to demonstrate their existence with 
 certainty. Undoubtedly there is a gelatinous hydrate of lime, for 
 which, through separation with absolute alcohol, I have found the 
 formula CaO + 5 H 2 O, which seems to indicate the existence of 
 a still higher hydrate. 
 
 In this way cement mortar will continue for years to expand, to 
 crack, and to destroy itself, each absorption of water causing 
 expansion, each drying out shrinkage ; each absorption of carbonic 
 acid causing either expansion or shrinkage, according to whether 
 simple or higher hydrates are changed ; and a substance as inelastic 
 as cement mortar can not, as a rule, resist for any length of time 
 the influence of such occurrences. All these drawbacks will clearly 
 be lessened if the caustic lime remaining or becoming free during 
 set can be bound by giving it a chance to form stable hydro-com- 
 pounds with hydraulic silica and alumina. 
 
 It is true that the lixiviation of the caustic alkalis can not be pre- 
 vented in this way, but on account of their small amount and their 
 ready solubility they are not of great importance and certainly do 
 not endanger the permanence of the mortar. In water they are 
 simply washed out, and when hardening takes place in air they 
 are a disfigurement more than anything else, and besides a tem- 
 porary one only, since, unlike the exudations of lime, they can be 
 removed by washing with water. 
 
 15671 4 
 
50 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 It is clear now that 110 greater mistake could be made than add- 
 ing lime paste to hydraulic binding media, which already contain 
 an excess of lime, that is, to Portland cement mortars; their 
 destruction is simply hastened in this way. My investigations 
 demonstrated clearly that even in mortars exposed to fresh water 
 the results of this proceeding are only apparently favorable, while 
 in reality they are injurious; the excess of lime dissolves rather 
 easily, and its washing out leaves the mortar more porous and 
 therefore more susceptible to disintegration. After two or three 
 years a considerable diminution of strength becomes apparent in 
 such cement-lime mortars. Frequently, it is true, carbonic acid 
 may exercise a favorable influence ; but I demonstrated long ago 
 that this proceeding is justified only in the case of poor mortars 
 exposed to air, in order to increase their plasticity. 
 
 In No. 33, 1882, 'of the "Deutsche Topfer und Ziegler Zeitung," 
 I discussed the question of the excess of lime in cement as fol- 
 lows: 
 
 When Portland cement is tempered with water there undoubtedly takes place 
 during set a shifting of molecules simultaneously with and in consequence of the 
 absorption of water. In the alkaline waters thus forming in the cement about 
 one -third of the existing lime separates in crystals of hydrate of lime during the 
 the process of hardening. This crystalline lime, instead of having any binding 
 power, has rather a tendency to destroy the cohesion obtained; but in good 
 cements this tendency will not be realized, because the very gradual separation 
 of lime finds cohesion too much advanced. 
 
 Considering this state of affairs, we may a priori infer that the quantity of 
 actual cement in the mortar can be increased by an addition of pozzuolana ; that 
 is, a substance with which hydrate of lime will combine to form cement ; in this 
 way the crystallizing of caustic lime can be prevented altogether, all hydrate of 
 lime, as it becomes free and before it has crystallized from the solution, being 
 employed in forming hydrosilicate or hydroaluminate of lime. 
 
 The "German Union of Cement Manufacturers" in 1882 opposed 
 this view in the following words : 
 
 Standard Portland cements do not stand in need of any so-called improving 
 admixture ; such admixtures cause a decrease of strength almost proportional 
 to their quantity. 
 
 In a published communication to the "Union," 1884, I again 
 demonstrated the correctness of my theoretical deductions concern- 
 ing hardening in fresh water. 
 
 The Royal testing station for binding media in Berlin has not been 
 able to elicit any thing but the absolute confirmation of my assertions, 
 and Professor L. von Tetmajer published in the " Schweizerische 
 Banzeitung," No. 24, 1884, his observations on the action of some 
 admixtures on Portland cement, showing an actual increase of 
 tensile and crushing strength, although the density of such admix- 
 tures was much less than that of Portland cement. The improve- 
 
INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 51 
 
 ment of Portland cement by means of suitable admixtures was 
 thus clearly demonstrated.* 
 
 The study of the behavior of hydraulic mortars in sea water 
 having convinced me that the principal cause of their frequent 
 failure is to be sought in the presence of free lime, it seemed clear 
 that such mortars could be improved by admixtures containing 
 hydraulic silica. By means of such suitable admixtures the entire 
 excess of lime could be bound and the mortar thus made very much 
 less susceptible to decomposition. Aside from permanence, it 
 seemed certain that this would very soon and unmistakably find 
 expression in the strength of the mortar, consequently the behavior 
 of mixed hydraulic binding media in sea water was necessarily the 
 most rigid criterion of my theory. I was thus induced to formu- 
 late a method of test for that purpose. 
 
 The action of sea water on these binding media is principally a 
 chemical one, and in order to obtain early results in this direction 
 it was necessary to exclude everything impeding such action. It 
 would have been a mistake to experiment with impervious mortars, 
 or with mortars protected by an outer shell through absorption of 
 carbonic acid; the object in view demanded rather the selection 
 of porous mortars, in which the action of the sea water would meet 
 with no obstruction. 
 
 The following solutions were employed for my experiments : 
 
 (1) Artificial sea water, composed as follows : 
 
 Kitchen salt.. grams. 30 
 
 Sulphate of magnesium do_. 12 
 
 Chloride of magnesium.-- do__ 3 
 
 Sulphate of calcium do . _ 1 
 
 Water liter. 1 
 
 Alkaline carbonate was left out purposely, there being no object 
 in weakening the action of the sea water. In order to keep the per- 
 centage of sulphuric acid constant, cylinders of gypsum wrapped 
 in linen were suspended in the water. The latter was renewed 
 daily in the beginning, after that weekly, and at the end of three 
 months monthly ; it was thoroughly stirred every day. 
 
 (2) Saturated solution of sulphate of calcium with cylinder of 
 gypsum suspended in it. 
 
 (3) One per cent solution of crystallized sulphate of magnesium. 
 
 (4) Two per cent solution of crystallized sulphate of magnesium. 
 
 (5) Three per cent solution of crystallized sulphate of magnesium. 
 
 (6) One and three-tenths per cent solution of crystallized sul- 
 phate of soda. 
 
 *See Zum Dogma, etc., 1884; again E. Dietrich Wochenblatt fur Baukunde,. 
 1885, Nos. 93 and 95, W. Michaelis, ibidem, 1888, No. 43. 
 
52 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 L. Vicat had already employed a solution of sulphate of soda, 
 and had recognized its destructive nature on hydraulic binding 
 media. 
 
 Fifteen years ago I found that a solution of sulphate of calcium 
 with only 0.127 per cent of sulphuric acid (SO 3 ) will completely 
 destroy neat Portland cement. I then perceived that only sulphuric 
 acid acts destructively on hydraulic binding media exposed to sea 
 water, and that magnesia is simply one of the visible symptoms 
 of such destruction, which, far from being injurious, rather causes 
 a favorable influence by closing the pores, thus restricting destruc- 
 tion to its beginning. 
 
 Still, it is readily understood why the action of sulphate of mag- 
 nesium must be a much more energetic one than that of sulphate 
 of calcium; the latter can only cause the formation of the double 
 compound sulphate and aluminate of lime, while the former, besides 
 this, will change the free and loosely-bound lime with which it comes 
 in contact into gypsum. The enormous expansion and destruction 
 often seen in cement mortar exposed to sea water can be understood 
 when we remember that at least one-third of the cement changes 
 into sulphate of calcium with 2 equivalents of water. 
 
 Portland cements rich in alumina are especially susceptible to 
 decomposition by a solution of sulphate of calcium, and most mari- 
 time structures built with these binding media have been saved 
 from complete and early ruin merely through such favorable cir- 
 cumstances as impermeability, absorption of carbonic acid, incrusta- 
 tion, closing of pores, etc. 
 
 I will now proceed to render an account of the experiments made 
 to support my assertions. In the main these experiments consisted 
 of two series, one of which has been completed, the other com- 
 menced. The first series consisted principally of rich mortars, viz, 
 1 : 2.5 and 1:4; the second series consisted of porous mortars, prin- 
 cipally in the proportion of 1:5. 
 
 For the first series, Bauschinger bars of 5 square centimeters 
 cross section and a length of 10 centimeters were made, the sand 
 used being, where not otherwise stated, quartz sand of mixed grain, 
 and the trass being Rhenish trass from the Nette valley. The 
 mortar was gauged in the following proportions by weight : 
 
 (1) Four of Portland cement (Stern brand) containing less than 
 6 per cent alumina and 10 of quartz sand. 
 
 (2) Two of same cement, 2 of trass, and 10 of quartz sand. 
 
 (3) Four of same cement, 1.2 of hydrate of silica (air-dried), and 
 13 of quartz sand. 
 
 (4) Four of same cement, 1.1 of kaolin from Zettlitz baked at 
 red heat, and 12.7 of quartz sand. 
 
INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 53 
 
 (5) Four of Portland cement, containing about 9 per cent alu- 
 mina (raw material washed), and 10 of quartz sand. 
 
 (6) Two of same cement, 2 of trass, and 10 of quartz sand. 
 
 (7) Four of same cement and 16 of standard sand. 
 
 (8) Four of same cement, 1.2 of hydrate of silica (air-dried), and 
 13 of quartz sand. 
 
 (9) Four of same cement, 1.1 of kaolin from Zettlitz baked at 
 red heat, and 12.7 of quartz sand. 
 
 (10) Four of Portland cement, containing about 9 per cent alu- 
 mina (worked dry), and 10 of quartz sand. 
 
 (11) Two of same cement, 2 of trass, and 10 of quartz sand. 
 
 (12) Four of same cement and 16 of standard sand. 
 
 (13) Four of same cement, 1.2 of hydrate of silica (air-dried), 
 and 13 of quartz sand. 
 
 (14) Four of same cement, 1.1 of kaolin from Zettlitz baked at 
 red heat, and 12.7 of quartz sand. 
 
 (15) One hundred of Bavarian Roman cement and 36 of water. 
 
 (16) Four of same Roman cement and 10 of quartz sand; 13 of 
 water. 
 
 (17) One hundred of Bosnian Roman cement and 36 of water. 
 
 (18) Four of same Roman cement and 10 of quartz sand; 13 of 
 water. 
 
 (19) Four of lime of Theil and 12 of standard sand. 
 
 (20) Four of lime of Theil and 20 of standard sand. 
 
 (21) Five of hydrate of silica (air-dried), 12.5 of paste of marble- 
 lime (32 per cent residuum after calcination), and 20 of quartz 
 sand. 
 
 (22) Six of kaolin from Zettlitz baked at red heat, 16.5 of paste 
 of marble-lime (32 per cent residuum after calcination), and 28.5 
 of quartz sand. 
 
 (23) Four of anhydrite of silica and 19 of paste of marble-lime 
 (32 per cent residuum after calcination). 
 
 All bars made from these mortars were allowed to harden in 
 moist air, but protected against carbonic acid ; after that, half of 
 the bars were exposed in hermetically-sealed bottles to a solution 
 of gypsum kept in a state of saturation, and the other half exposed 
 in the same manner to a 1 per cent solution of sulphate of mag- 
 nesium. 
 
 The solution was renewed daily during the first two weeks, then 
 weekly, and at the expiration of three months monthly. 
 
54 
 
 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 The following observations were made on the bars exposed to 
 the solution of gypsum : 
 
 Mortar No. 
 
 Commencement of destruction. 
 
 Total destruction. 
 
 1 
 
 After 6 months. 
 
 After 1 year. 
 
 5 
 
 After 6 months 
 
 After 1 year. 
 
 7 
 
 After 3 months 
 
 After 6 months. 
 
 10 
 
 After 3 months 
 
 After 9 months. 
 
 12 
 
 After 3 months 
 
 After 6 months. 
 
 20 
 
 After 1 year 
 
 
 22 
 
 After 14 days 
 
 After 1 month. 
 
 
 
 
 All others are intact at this date, that is, after a lapse of two 
 years. 
 
 The following observations were made on the bars exposed to a 
 1 per cent solution of sulphate of magnesium : 
 
 Mortar No. 
 
 Commencement of destruction. 
 
 Total destruction. 
 
 1 
 
 After 6 months 
 
 After 1 year. 
 
 5 
 
 After 6 months 
 
 After 9 months. 
 
 7 
 
 After 4 months 
 
 After 6 months. 
 
 10 
 
 After 3 months 
 
 After 9 months. 
 
 12 
 
 After 3 months 
 
 After 6 months. 
 
 17.. 
 
 After 18 months . 
 
 
 20 . 
 
 After 12 months 
 
 
 22 
 
 After 8 days 
 
 After 1 month. 
 
 
 
 
 These tests demonstrate that mortar No. 22, consisting of kaolin 
 and paste of lime, is rapidly destroyed in a solution of gypsum, as 
 well as in one of sulphate of magnesium, while the same mortar 
 remained perfectly sound during exposure to fresh water. There 
 are two reasons for this : In the first place, kaolin baked at red 
 heat is a very porous pozzuolana, and therefore renders a mortar 
 of little density; in the second place, kaolin is richer in alumina 
 than any other pozzuolana, being composed of 55 per cent of silica 
 to 42 per cent of alumina. 
 
 Mortar No. 22, therefore, was composed as follows : 
 
 3. 30 parts by weight of silica or, in equivalents, 550 ; 
 
 2.52 parts by weight of alumina or, in equivalents, 244; 
 
 5.28 parts by weight of lime or, in equivalents, 943. 
 
 It is seen that the almost exact formation of SiO 2 CaO -f- A1 2 O 3 , 
 3 CaO is possible here ; the aluminate is the sole cause of destruc- 
 tion, and the rapidity of the latter is due to the high percentage of 
 alumina. We have seen above that 1 part by weight of alumina 
 can form nearly 12 times as much of aluminate and sulphate of 
 lime by combining with 3C equivalents of water ; now mortar No. 
 
INFLUENCE OF SEA WATER OX HYDRAULIC MORTARS. 
 
 00 
 
 22 contains 22.7 per cent of alumina, consequently 270 parts by 
 weight of the double compound may be found in it. 
 
 Notwithstanding this, the three Portland cement mortars mixed 
 with kaolin have stood well so far (for 2 years), doubtless owing 
 to the combining of the free lime with the admixture. 
 
 The unmixed Portland cements have decayed rapidly ; those with 
 9 per cent of alumina, in accordance with theory, sooner than the 
 Stern cement (containing only 5 to 6 per cent). The lime of Theil 
 has stood so far; the porous mortar of 1 : 5, the hardening of which 
 took place under exclusion of carbonic acid, shows commencement 
 of destruction at the most sensitive part of one bar, viz, the edge. 
 
 The Bavarian Roman cement, both neat and mixed with quartz 
 sand in the ratio of 1:2.5, has so far remained perfectly intact. 
 The Bosnian Roman cement, in solution of sulphate of magnesium, 
 showed commencement of destruction on one of the edges after 18 
 months; the sand mortars of the same cement are still intact. 
 The composition of these two Roman cements when thoroughly 
 baked is as follows : 
 
 
 Bavarian. 
 
 Bosnian. 
 
 Sand and clay 
 
 Per cent. 
 
 6.381 
 23. 881 
 9.709 
 4.052 
 47. 229 
 3.992 
 4.237 
 
 Not det< 
 
 Per cent. 
 
 2.920 
 30. 180 
 9.036 
 3.669 
 49.000 
 2.215 
 2.109 
 
 3rmined. 
 
 Silica 
 
 Alumina - 
 
 Ferric oxide 
 
 Lime 
 
 Magnesia -- 
 
 Sulphuric acid 
 
 Manganese ) 
 
 Alkali \ 
 
 
 99. 481 
 
 99. 129 
 
 The most highly calcareous compounds mentioned in the begin- 
 ning of this article can not, as has been shown above, form in these 
 two cements ; if the formation of 2 R 2 O 3 , 3 CaO is assumed, then 
 for saturation of the silica there remain in the Bavarian cements 
 1.5 equivalents of lime, that is, the compound 2SiO 2 , 3 CaO, and 
 in the Bosnian cement 1.35 equivalents of lime, that, is, the com- 
 pound 3SiO.,, 4 CaO. Hence the great resistance of the Roman 
 cements to the action of sea water. 
 
 Long before the appearance of any visible symptoms of destruc- 
 tion, the existence or nonexistence of injury will become manifest 
 in tests of strength, especially in comparative tests of test pieces 
 hardened in freshwater and others hardened in sea water, and 
 such tests evidently form the correct method of determining the 
 influence of sea water. 
 
56 
 
 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 Test pieces having a great surface area in proportion to the cross 
 section, to be tested clearly, are best adapted for such tests. Besides 
 the regular tension briquettes I have employed bars 30 centimeters 
 long, 2 centimeters high, and 4 centimeters broad. These bars were 
 tested as to their transverse strength by centerloads over a clear 
 span of 21.33 centimeters. The well known equation for the coeffi- 
 
 cient of resistance, C - 
 
 - , , therefore, renders this coefficient 
 
 (per square centimeter) equal to 2 W, that is, twice the breaking load. 
 As mentioned above, it is not advisable to employ impervious 
 mortars for testing the influence of sea water on hydraulic binding 
 media, because the time required for obtaining results would be 
 unduly increased, and because it is evident that methods of test 
 are the better the sooner they lead to results. Therefore, mortars 
 of 1 : 5 are much preferable to mortars of 1:3. 
 
 In accordance with this line of reasoning I have undertaken the 
 second series of tests, which is not yet completed. The Stettin 
 Portland cement (A), the same Portland cement worked dry con- 
 taining 9 per cent alumina (B), the lime of Theil (C), and the 
 Bavarian Roman cement (D) are the same as those employed for 
 the first series of tests. The composition of these cements in a 
 baked condition is as follows : 
 
 
 A 
 
 B 
 
 c 
 
 D 
 
 Silica 
 
 Per cent. 
 21.712 
 
 Per cent. 
 
 21 331 
 
 Per cent. 
 
 23 880 
 
 Per cent. 
 
 23 881 
 
 Alumina 
 
 5.805 
 
 8.918 
 
 2.570 
 
 9.709 
 
 Ferric oxide 
 
 2.949 
 
 2.695 
 
 0.880 
 
 4 052 
 
 Lime _. 
 
 64. 851 
 
 63. 776 
 
 69 150 
 
 44 263 
 
 Magnesia 
 
 1.030 
 
 805 
 
 1 600 
 
 3 992 
 
 Potash 
 
 0.748 
 
 777 
 
 140 
 
 
 Soda - . 
 
 0.160 
 
 0.101 
 
 0.070 
 
 
 Sulphate of lime 
 
 2.468 
 
 1.631 
 
 1 090 
 
 7 203 
 
 Residuum 
 
 0.357 
 
 0.123 
 
 
 6 381 
 
 
 
 
 
 
 
 100. 080 
 
 100. 157 
 
 99. 380 
 
 99.481 
 
 Standard tensile briquettes were shaped from mortars mixed in 
 the proportion of 1 : 5 with Berlin standard sand. 
 
 In the place of sea water the artificial sea water described above 
 was employed. Unless otherwise stated the test pieces were first 
 allowed to harden in moist air for 24 hours. Those under S were 
 stored on edge in sea water, and those under R in fresh water ; the 
 additional index "e" signifies that before being submerged in sea 
 water the briquettes were kept in air for 8 weeks (being moist- 
 ened daily) to allow the absorption of carbonic acid. 
 
 The test pieces made of lime of Theil were first kept in moist 
 air for 28 days and then 8 weeks in air, being moistened daily. 
 
INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 57 
 
 Tlie time of test is counted from the day previous to immersion in 
 water. At the present writing the test pieces have been under 
 treatment for 20 months. 
 
 Each of the following figures is the mean of ten individual tests, 
 and represents kilograms per square centimeter : 
 
 Age since 
 immersion. 
 
 A i; 
 
 A S 
 
 ASe 
 
 BR 
 
 BS BSe 
 
 CB 
 
 CS 
 
 CSe 
 
 DB 
 
 DS DSe 
 
 7 days 
 
 28 days 
 
 90 days 
 
 7.23 
 10.09 
 11.60 
 
 1 year _ 16 
 
 7.54 
 10.40 
 16 
 
 15.93 
 15.25 
 15.28 
 6. 50 to 18 
 
 10.50 7.86 j 17.05 
 12.68 j 6.91 I 17.75 
 15 i 9.10 j 15.89 
 16.70 11.20 2 to 15 
 
 3.81 
 5.17 
 
 8.38 
 13.25 
 
 4.96 
 6.50 
 10.86 
 15.20 
 
 7.75 
 6.97 
 
 12.50 
 
 9.22 
 
 2.86 
 
 5.11 
 
 9.68 I 11.43 
 
 14.43 I 14.12 
 
 6.27 13. 
 
 12 
 13.50 
 14.66 
 15.44 
 
 Cement A was then mixed with trass, the following being taken 
 into consideration: 
 
 100 parts by weight of trass contain 
 
 10 to 12 parts of water and loss in baking; 
 
 20 to 30 parts of hydraulic silica and alumina ; 
 
 60 to 65 parts of minerals acting as sand. 
 
 Of the trass employed 41 per cent was held by a sieve of 2,500 
 meshes per square centimeter ; in an air-dried condition its compo- 
 sition was as follows : 
 
 Percent. 
 
 Hygroscopic water 4. 141 
 
 Water of crystallization 6. 899 
 
 Lossat900C 0.202 
 
 Silica 53.583 
 
 Alumina 19. 008 
 
 Oxide of manganese* 0. 115 
 
 Ferricoxide* 4.193 
 
 Lime 1.736 
 
 Magnesia 1. 652 
 
 Potash 4.147 
 
 Soda 4.242 
 
 Titanic acid \ 
 
 Chlorine > Not determined. 
 
 Phosphoric acid. ) 
 
 Sulphuric acid 0. 107 
 
 100. 025 
 With a 10 per cent lye of caustic soda this trass rendered: 
 
 (a) Digested for 10 hours in a water bath p er cent. 
 
 Hydraulic silica 16.543 
 
 Hydraulic alumina 4. 810 
 
 (6) Digested for 24 hours in a water bath 
 
 Hydraulic silica 16. 708 
 
 Hydraulic alumina 6. 043 
 
 *Iroii and manganese were counted as oxides, although there also existed 
 protoxides of them. 
 
INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 The following two mortars were made : 
 
 E. 
 
 1 part cement A... ) 1 25 bin ding medium , that is, somewhat richer 
 
 1 part trass ) 
 
 4 parts standard sand 
 
 1 
 
 0.5 
 ( 4. 5 parts standard sand 
 
 4. 60 sand 
 
 than 1:4. 
 
 25 binding medium * that is > somewhat 
 
 * sand ---------- f poorer than 1 ; 4. 
 
 In the following table, which renders the results of tests of 
 strength with these mortars, the letters R and S have the same 
 meaning as above; each of the figures is the mean of ten indi- 
 vidual tests: 
 
 Age. 
 
 E K 
 
 E 3 
 
 F R 
 
 . F S 
 
 7 days 
 
 9.80 
 
 11.80 
 
 11.05 
 
 10.10 
 
 28 days 
 
 19.15 
 
 28.00 
 
 16.90 
 
 19.55 
 
 90 days 
 
 26.70 
 
 35.70 
 
 21.80 
 
 23.65 
 
 1 year 
 
 30.95 
 
 39.50 
 
 27.55 
 
 24. 59 
 
 
 
 
 
 
 It has been shown that out of 100 parts by weight of Portland 
 cement, with a mean percentage of lime, there will separate 25 
 parts of CaO (33 parts of CaH 2 O 2 ). If 100 parts of trass are 
 added (which, for example, contain 16.5 parts of silica and 5.14 
 parts of alumina, both of them able to combine) the formation of 
 SiO 2 CaO would require 15.4 parts of CaO, and the formation of 
 2 A1 2 O 3 , 3 CaO would require 4.2 parts of CaO, that is, a total of 
 19.6 parts of CaO; therefore the formation of a silicate richer in 
 lime than the monosilicate is possible, and an admixture of 125 
 parts or more of trass per 100 parts of Portland cement is very 
 probably advisable, all the more so since compounds, even of 
 2 SiO 2 to 1 or 2 of CaO, become very hard and are doubtless more 
 stable than the monosilicate. At any rate the admixture should 
 be the greater the higher the percentage of lime in the cement ; 
 this holds good, likewise, for hydraulic limes. 
 
 In accordance with this an injurious influence of sea water is 
 perceptible in mortar F S after the lapse of a year ; in this mixture 
 of 100 parts of cement and 50 of trass, only 10 parts of the free lime 
 will be bound with sufficient permanence, while 15 parts of lime 
 will remain free to enter new compounds, and even were the sili- 
 cate 2 SiO 2 3 CaO and the aluminate 2 Al a O 8 5 CaO to form, there 
 still would remain 10 parts by weight of lime free to be acted 
 upon by the sulphates of the sea water. 
 
 Later experiments, commenced in October, 1895, refer to mortars 
 composed of Portland cement and lime of Theil, with admixtures 
 of trass in such proportion that the richness of the mortar is rather 
 
INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 59 
 
 diminished than increased by them. These mortars are composed 
 as follows : 
 
 G. Portland cement with 9 per cent alumina (as above: B) 1 : 5 
 
 Berlin standard sand. 
 
 H. Same cement 1 part, trass 1 part, Berlin standard sand 6 parts. 
 I. Same cement 1 part, trass 1 part, Berlin standard sand 6. 75 parts. 
 K. Lime of Theil (as above : C) 1 : 5 Berlin standard sand. 
 L. Lime of Theil 1 part, trass 1 part, Berlin standard sand 5 parts. 
 M. Lime of Theil 1 part, trass 1 part, Berlin standard sand 6 parts. 
 
 The ratio of binding medium to sand is as follows : 
 
 H. 1 part by weight of binding medium to nearly 5. 3 parts of sand. 
 I. 1 part by weight of binding medium to nearly 6 parts of sand. 
 L. 1 part by weight of binding medium to nearly 5 parts of sand. 
 M. 1 part by weight of binding medium to nearly 5.6 parts of sand. 
 
 K, L, and M were first allowed to harden in moist air for 7 days. 
 
 The tests of mortars E to M are to be extended over a period of 
 three years. The broken briquettes of one year's age and above 
 will be put partly into a 2 per cent, partly into a 3 per cent, solu- 
 tion of sulphate of magnesium, being thus exposed to much more 
 energetic action than that of the strongest sea water. 
 
 In the following table there are rendered the results of tests of 
 strength so far attained ; each result is the mean of 10 individual 
 tests, and is expressed in kilograms per square centimeter ; the des- 
 ignations R and S have the same meaning as heretofore ; opposite A 
 is rendered the specific gravity of mortars immediately preceding 
 the tests : 
 
 Age since immersion. 
 
 GR 
 
 GS 
 
 HR 
 
 HS 
 
 IR 
 
 IS 
 
 KS 
 
 LS 
 
 MS 
 
 28 days 
 
 8.80 
 
 6.35 
 
 10.40 
 
 21 65 
 
 10 10 
 
 20 45 
 
 2 97 
 
 11 75 
 
 10 30 
 
 
 
 
 
 
 
 
 
 2.162 
 
 2 195 
 
 2.227 
 
 90 days 
 
 10.25 
 
 7.55 
 
 16.85 
 
 24.55 
 
 16.55 
 
 24.60 
 
 2.3d* 
 
 21.57 
 
 20.10 
 
 ^ 
 
 2.137 
 
 2.172 
 
 2.26 
 
 2.28 
 
 2.274 
 
 2.292 
 
 2 252 
 
 2 250 
 
 2.259 
 
 
 
 
 
 
 
 
 
 
 
 * Besides a decrease of strength there was also perceptible in the 90-day test pieces of lime of 
 Theil an expansion of the outer shell here and there. 
 
 It is at once apparent from these figures that in all mortars con- 
 taining lime which becomes free during the hardening process, sea 
 water exercises an influence opposed to an increase of strength. 
 
 The process of crystallization and the process of hardening strug- 
 gle with each other in these cases, the former generally being vic- 
 torious and causing complete destruction of the cohesion obtained 
 through the latter. 
 
 On one of the test pieces B S symptoms of decay were visible 
 after 90 days ; it was cracked and decomposed to such an extent 
 that it broke in putting it into the testing machine ; on the out- 
 line of the fracture a deposit of hydrate of magnesia to a depth 
 of some millimeters was visible. At the end of a year the deposit 
 
60 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 of magnesia was found in all test pieces but one to have advanced 
 to a depth of 5 millimeters ; the same condition was found to exist 
 in B Se. 
 
 Strange to say, the destruction of the test pieces that had been 
 exposed to absorption of carbonic acid was even a more rapid one,* 
 the carbonated shell being lifted and rolled up, as in A Se as com- 
 pared with A S, and in C Se as compared with C S, all at an age of 
 one year ; D Se is an exception ; all test pieces made of mortar D 
 were entirely sound, and did not show any deposit of magnesia at 
 the contours of fracture. 
 
 The above figures speak for themselves ; an admixture of trass 
 (or, generally, of effective pozzuolanas) to hydraulic binding media 
 that are too rich in lime (Portland cement and hydraulic limes), 
 may increase two or three fold the strength of mortars made from 
 them, and may make these mortars stable in sea water. The causes 
 of increase of strength are evident; best pozzuolanas contain a 
 quantity of effective hydraulic factors equivalent to those in the 
 best Portland cement, \ and as Portland cement contains a sufficient 
 excess of lime to satisfy fully the pozzuolanas, it is not surprising 
 that from a combination of both materials mortars of increased 
 strength should result. 
 
 Clearly admixtures containing much silica and little alumina are 
 preferable ; hence an admixture of kaolin, in the case of maritime 
 structures, is not advisable. 
 
 Summiiig up, it may therefore be asserted that scientific and 
 practical proof has been rendered of the nonsuitability for maritime 
 structures of hydraulic binding media containing more lime than is 
 required for the formation of stable hydrosilicates and aluminates 
 (these compounds are stable only when their percentage of lime is 
 small, and the smaller it is the greater will be their stability). 
 
 Mortars mixed according to my suggestion, besides being much 
 more stable and stronger, are also considerably cheaper, and it is 
 therefore to the public interest that they should be employed 
 generally for maritime structures, and the use of excessively cal- 
 careous hydraulic binding media without such proper admixtures 
 should be discontinued. 
 
 * An explanation of this fact may perhaps be found in the increased porosity 
 of the mortar through absorption of carbonic acid and the greater ease of access 
 thus afforded to the saline solution. The hydrosilicate of lime certainly is in a 
 colloidal condition. When decomposed by carbonic acid crystalline carbonate 
 of lime will form and the silica will separate ; but as compared with a colloidal 
 condition this process of carbonation must render a mortar of much inferior 
 density. 
 
 f According to my latest experiments genuine trass contains nearly 50 per cent 
 of silica and alumina able to enter new compounds. 
 
INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 61 
 
 This solution of the problem is, indeed, the most favorable one 
 that can be conceived. No changes are introduced beyond the 
 judicious use of such well-approved binding media as the true 
 pozzuolanas, among which none is superior to genuine trass, that 
 is, finely-ground hydraulic tufa stone, without any extraneous 
 matter whatsoever. But in view of the strength and the energy 
 of set of Portland cement (with which the mixture is made), it is 
 quite admissible to employ not only trass, but also the pozzuolanas 
 endowed with less vigor of set. Theoretical computation as well 
 as practical tests, which can be made rapidly, will in each case 
 show which mixture is the one best adapted to the circum- 
 stances. * * * 
 
 It is not my business to investigate how consumers will look 
 upon this matter, but I still adhere to the view that the manufac- 
 turer is not only the best judge of the admixtures required by 
 his particular product, but also best able to carry out the particu- 
 lar work of mixing. * * * But this point should be decided 
 by the consumers themselves. 
 
 This discussion would doubtless have been of greater value if its 
 publication had been postponed for some years, but my advanced 
 age, as well as the consideration of a number of great maritime 
 structures soon to be commenced in various countries, has induced 
 me to communicate my results now. May my work be repeated, 
 tested, and carried on to completion by men of science and by 
 practical men. 
 
 # * * # * * * 
 
 My present suggestion is based on the fact, long ago asserted by 
 me, that hydraulic limes and Portland cements overloaded with 
 lime can be improved by giving them such admixtures, either 
 during the process of manufacture or during the gauging of the 
 mortar. 
 
 The following pozzuolanas are suitable for this purpose : Hydrau- 
 lic silica per se, opal, infusorial earth, the pozzuolanas proper, 
 trass, santorin earth, powdered glass, burnt alum slate, kaolin, 
 brick powder, etc. 
 
 Of all such admixtures, so far known, genuine trass is the most 
 effective. 
 
 I have made the following propositions to the permanent com- 
 mittee in regard to testing hydraulic binding media as to their 
 resistance against sea water : 
 
 The test of resistance against sea water is made with porous 
 mortar of 1:5, briquettes being kept in artificial or in natural 
 sea water, which is renewed daily during the first four weeks and 
 after that weekly ; the water must be stirred daily ; tests may be 
 made of tensile crushing or transverse strength. 
 
62 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 Artificial sea water is a solution of the following ingredients in 
 1 liter of water : 
 
 Grams. 
 
 Kitchen salt 30 
 
 Chloride of magnesium 3 
 
 Sulphate of magnesium 3 
 
 Sulphate of calcium 1.25 
 
 No bicarbonates are added, the object not being to retard the 
 action of sea water; but bars of gypsum wrapped in linen are 
 immersed in the artificial sea water in order to keep its percentage 
 of sulphuric acid constant. 
 
 REASONS. The action of sea water on hydraulic binding media 
 is principally a chemical one, resulting from the action upon each 
 other of the soluble sulphates contained in sea water, and the free 
 lime, that is, the lime becoming free during set ; the ferrate, the 
 aluminate, and the silicate of lime. 
 
 This action will become apparent soonest in porous mortar, tests 
 of strength giving the earliest evidence of it. Therefore test 
 pieces are recommended having a large superficial area as com- 
 pared with the area of fracture, and consisting of mortar gauged 
 in the proportion of 1 : 5. Tensile and transverse tests are prefer- 
 able to crushing tests. 
 
 Comparative tests of briquettes hardened in sea water, and others 
 hardened in fresh water, will very soon enable the experimenter 
 to form an opinion as to the influence of sea water in any particu- 
 lar case. A 2 per cent solution of sulphate of magnesium will lead 
 to results in still less time. 
 
 The idea of the action of salts of magnesium on hydraulic 
 mortars should be abandoned, magnesia not acting in an injurious 
 manner, but rather in a useful one, through closing of pores ; it 
 separates in the form of a soft, spongy hydrate, which does not 
 cause any strains, and, being insoluble, remains at the place of 
 formation. Sulphuric acid only acts destructively, and therefore 
 all soluble sulphates act so likewise (although with different in- 
 tensity), whether they be sulphate of soda (Vicat), or sulphate of 
 calcium (Michaelis), or sulphate of magnesium. 
 
REPLY TO DR. MICHAELIS'S ARTICLE.* 
 
 BY THE BOARD OF DIRECTORS OF THE UNION OF GERMAN PORTLAND CEMENT 
 
 MANUFACTURERS. 
 
 In reply to the foregoing article we desire first to give the fol- 
 lowing historical data : 
 
 When in the beginning of the eighties the mixing of Portland 
 cement with slag and other inferior material threatened to become 
 too prevalent, the board of directors of the German Portland Cement 
 Manufacturers felt called upon energetically to oppose such pro- 
 ceedings, because it was contrary to usage and justice to designate a 
 mixture of Portland cement with other materials simply as ' ' Port- 
 land Cement," and secondly, because the admixtures employed, 
 especially slag, diminished the quality of the product, and the ever- 
 increasing extent of the practice was thus calculated seriously to 
 injure the reputation of the German cement industry. 
 
 The manufacturers then attempted to prove that an admixture 
 of slag, so far from depreciating, actually improved the quality of 
 cement by increasing its strength. In support of that assertion, 
 Dr. Michaelis's theory was quoted, according to which pozzuolanas 
 (that is, compounds containing silica and alumina able to combine) 
 can combine with the lime becoming free during set of Portland 
 cement, forming with it a new cement-like compound, thus increas- 
 ing the strength through better cementation of the particles. It is 
 not to be denied that the idea forming the basis of this theory is a 
 correct one, inasmuch as limited percentages of certain admixtures 
 for instance, ultramarine and powdered hydrate of silica will 
 increase the strength of Portland cement. But for the pozzuo- 
 lanas in the market the same can not be said. For instance, admix- 
 tures of trass and of granulated slag will not cause an increase of 
 strength. But slag was quite frequently used as an admixture, and 
 
 *The controversy between Dr. Michaelis and the Union of German Cement 
 Manufacturers was not limited to the technical question itself, but led to a dis- 
 cussion of motives, which assumed a somewhat personal character and had no 
 bearing on the real point at issue. For these reasons this part of the discussion 
 has been omitted from the above translation, the omissions in each case being 
 designated in the customary way. 
 
 (63) 
 
64 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 it is apparent from the following table that the strength of cement 
 was thereby diminished : 
 
 TABLE No. 1. 
 
 [1 cement, 3 standard sand. Tensile strength in kilograms per square centimeter, after 28 days 1 hardening 
 
 in fresh water.] 
 
 Composition of cement. 
 
 Without U|tra . 
 
 s,t. 
 
 Trass. 
 
 Slag. 
 
 Fine sand. 
 
 Carbonate 
 of lime. 
 
 Hydrate 
 of lime. 
 
 
 20.8 
 
 
 
 
 
 
 90 per cent cement _) 
 
 
 
 
 
 
 
 
 23.6 
 
 20.4 
 
 18.4 
 
 18.2 
 
 18.2 
 
 19.0 
 
 10 per cent admixture j 
 80 per cent cement [ 
 
 
 
 
 
 
 
 
 24.5 
 
 18.1 
 
 15.4 
 
 15.7 
 
 16.1 
 
 15.1 
 
 20 per cent admixture j 
 
 
 
 
 
 
 
 
 _ _ 20. 3 
 
 15.7 
 
 13. 5 
 
 13.9 
 
 13.6 
 
 10.2 
 
 33 per cent admixture j 
 50 per cent cement ") 
 
 17.1 
 
 12.5 
 
 10.2 
 
 11.0 
 
 10.4 
 
 
 50 per cent admixture j 
 
 
 
 
 
 
 
 If these admixtures really had turned out to be improvements 
 the mixed cements would doubtless have prevailed. Instead of 
 that, the practice of mixing has decreased more and more. * * * 
 It may also be stated here that up to the present day no practi- 
 cally applicable material is known, which, as an admixture with 
 Portland cement, will increase its strength in air or in water. 
 
 Passing now to a review of Dr. Michaelis's paper we shall first 
 discuss his theory of the set of Portland cement. Michaelis says : 
 It may be assumed that during hydraulic set the following compounds are 
 formed : 
 
 2SiO 2 , 3CaO + xH 2 O; 
 2Fe a O 8 , 4CaO + yH 2 O; 
 2A1 2 O 3 , 5CaO + zH 2 O. 
 
 By means of these formula and an analysis of Portland cement 
 he computes that a considerable portion of the lime contained in 
 Portland cement remains free after set ; he asserts that this free 
 lime separates in form of crystals to the amount of about 25 per 
 cent in an average Portland cement, and that it exercises an 
 injurious influence, especially in sea water, where through chemi- 
 cal changes it will cause destruction. 
 
 But Dr. Michaelis has nowhere furnished proof that the above 
 three compounds of lime did form during his experiments, or 
 that they actually do form during the process of hydraulic set of 
 Portland cement. He has likewise failed to prove that the forma- 
 tion of compounds with a higher percentage of lime than the 
 above-named ones is impossible during the hardening of Portland 
 cement. Consequently his assertion that 25 per cent of lime sepa- 
 rates during set of Portland cement is not established. But even 
 
INFLUENCE OF SEA WATER OX HYDRAULIC MORTARS. 65 
 
 supposing it were, this would not justify the conclusion that this 
 25 per cent of lime is injurious. As a matter of experience, Port- 
 land cement, if suitably employed, is the very one of all binding 
 media which has given more satisfaction than any other in air, 
 fresh water, and sea water. The high percentage of lime con- 
 tained in Portland cement is one of its characteristic features ; it 
 enables us to burn this cement up to vitrification, and thus endow 
 it with great density and those other excellent qualities which dis- 
 tinguish Portland cement so favorably from all other binding 
 media. 
 
 We can not here enter upon all the assertions made by Dr. 
 Michaelis, but must restrict ourselves to a discussion of his prin- 
 cipal theme, that is, the behavior of Portland cement in sea water, 
 and his suggestions for the improvement of sea- water mortars. 
 
 During the hardening of Portland cement in sea water the 
 hydrate of lime being then formed is acted upon principally by 
 chloride of magnesium and sulphate of magnesium (besides smaller 
 amounts of other sulphates). The chloride of magnesium changes 
 into chloride of calcium and hydrate of magnesia ; the former is 
 dissolved, while the latter, being insoluble, remains in the mortar. 
 A portion of the hydrate of lime, which under ordinary circum- 
 stances serves for better cementation of the mortar, thus increas- 
 ing its strength, is therefore lost in sea water. 
 
 The sulphate of magnesia combines with the hydrate of lime 
 contained in Portland cement, forming sulphate of calcium and 
 hydrate of magnesia. If the sulphate of calcium as such, and 
 more especially a double compound of it, viz, sulphate of alumina 
 and lime, could form in sufficient quantities, they would indeed 
 endanger the cement mortar, as both of these compounds, and 
 especially the last-named soluble compound, in expanding consid- 
 erably, absorb a great deal of water, and may thus destroy 
 cohesion. Continued action of the sulphates contained in sea 
 water is, however, soon energetically opposed by the increase of 
 closeness of grain taking place during hardening. In this way, 
 and by magnesia separating in the pores, the infiltration of sea 
 water is gradually checked and its action stopped. The correct- 
 ness of this view has been shown by R. Dyckerhoff's experiments, 
 according to which, absorption of sulphuric acid by Portland 
 cement mortar hardening in sea water diminishes as time passes, 
 rich cement mortars absorbing less and therefore suffering less 
 than poor mortars. 
 
 The above-described chemical processes furnish an explanation 
 of the fact that the strength of Portland cement mortar is less in 
 sea water than in fresh water. 
 
 15671 5 
 
06 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 But from Dr. Michaelis's paper it would appear that Portland 
 cement will not last at all in sea water. 
 
 The best criterion of a building material is the experience gained 
 during its practical use on a large scale. 
 
 Turning to practice, therefore, we find numerous maritime con- 
 structions on the shores of the German Ocean and of the Baltic 
 furnishing proof of the durability of Portland cements during long 
 periods. The seacoast forts of Copenhagen, for instance, which 
 were built nearly forty years ago, are to this day in a state of per- 
 fect preservation. We intentionally name these forts, as they were 
 built with Stettin cement of the same chemical composition as 
 cement A, which was used by Dr. Michaelis in the experiments 
 leading him to the conclusion that Portland cement will not last in 
 sea water. 
 
 We also wish to point to a publication made in 1889 by a com- 
 mission on limes, cements, and mortars, which was appointed by 
 the French ministry of public works. This commission has made 
 experiments extending over a period of ten years, on hydraulic 
 limes, on Roman and Portland cements, on blocks of concrete as 
 well as of masonry immersed in sea water, and has found that none 
 of the hydraulic limes under test resisted the action of sea water 
 longer than from four to five years. Even lime of Theil did not 
 resist much longer, at least not when exposed to agitated sea water, 
 although great power of resistance against sea water is frequently 
 claimed for that lime. Blocks formed of Roman cement were like- 
 wise found to have been injuriously affected after from eight to 
 ten years, while the Portland cement blocks were the only ones 
 that remained intact after a lapse of ten years. 
 
 On the other hand, it can not be denied that there have been fail- 
 ures of Portland cement in maritime structures. It is, however, sus- 
 ceptible of proof that in all such cases there have been either errors 
 of construction, or the mortar employed has been too poor and there- 
 fore too porous. For failures arising from conditions of this kind, 
 Portland cement, as such, can not be held responsible. * * * 
 
 Tests now in progress at Sylt, under the auspices of the Royal 
 Department of Public Buildings, have so far resulted in all bri- 
 quettes of Portland cement exposed to sea water having remained 
 intact and continued to increase in strength, although not as much 
 as in fresh water, while mortars of Portland cement and lime, and 
 especially mortars of trass and lime, have been injuriously affected 
 by sea water. 
 
 Dr. Michaelis recommends adding trass to Portland cement in 
 order to increase its power of resistance against the chemical action 
 of sea water, his theory being that the lime becoming free during 
 
INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 67 
 
 set will combine with the silica and alumina of the admixture, 
 forming a chemically stable compound. 
 
 In a series of tests executed by him he has obtained through an 
 admixture of trass a considerable increase of strength, especially 
 when hardening took place in sea water. 
 
 We desire here to call the reader's attention to the difference 
 between the practice of giving Portland cement an admixture, and 
 the practice of mixing into the mortar some additional material 
 besides Portland cement and sand. If a part of the cement is 
 replaced by other material (excepting ultramarine and similar 
 compounds), then according to Table No. 1 the mortar's strength 
 is diminished. 
 
 But if limited quantities of finely-ground material, such as 
 hydrate of lime, powdered sand, etc. , are added to a poor cement 
 mortar, then a filling of pores takes place, which will cause an 
 increase of strength. Evidently this is not an improvement of 
 cement, but of mortar. In the case of water storage especially, 
 trass acts better than other materials, because, aside from the fill- 
 ing of pores, a further increase of density takes place through the 
 trass combining with hydrate of lime separating from the 
 cement. 
 
 In rich mortars, however, that are less porous, no improvement 
 will be effected by the same admixtures. Even trass, notwith- 
 standing its free silica, will not improve them, the greater volume 
 of water absorbed by it causing inferior density, without any addi- 
 tional strength of cohesion between the particles. 
 
 For the purpose of testing the action of pozzuolanas during set 
 of cement mortar in sea water, we have commenced, like Dr. 
 Michaelis, several series of tests with cement mortars having an 
 admixture of trass. 
 
 For these tests we have selected principally rich mortars, because 
 these are mostly used for maritime structures. But for the pur- 
 pose of comparison there was also tested one poor mortar (1:4). 
 The storage water used was sea water from the German Ocean. 
 
 The cement employed, when tested according to standard method 
 (1 : 3, 28 days), was found to possess a tensile strength of 22.4 kilo- 
 grams and a crushing strength of 238.8 kilograms per square 
 centimeter. The trass used was from Plaidt ; the sand was quartz 
 sand, passing through a sieve of 60 meshes per square centimeter 
 and held by one of 900 meshes per square centimeter. All mortars 
 were of equal consistency. 
 
68 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 In the following table are rendered the results obtained thus far 
 
 TABLE No. 2. 
 
 TENSILE STRENGTH AFTER 28 DAYS. 
 
 
 1 
 
 2 
 
 3 
 
 4 
 
 5 
 
 6 
 
 7 
 
 8 
 
 Kind of storage. 
 
 1 cement, 
 1 sand. 
 
 1 cement, 
 1 trass. 
 
 1 cement, 
 1 trass, 
 1 sand. 
 
 1 cement, 
 2 sand. 
 
 1 cement, 
 1 trass, 
 2 sand. 
 
 1 cement, 
 4 sand. 
 
 1 cement, 
 1 trass, 
 4 sand. 
 
 ly cement, 
 4 sand. 
 
 Fresh water 
 
 29.9 
 
 29.3 
 
 28.5 
 
 23.5 
 
 19.3 
 
 11.5 
 
 15.8 
 
 16.2 
 
 Sea water 
 
 29.3 
 
 28.1 
 
 ,77 
 
 22.4 
 
 21. G 
 
 10.6 
 
 15.2 
 
 14.8 
 
 
 
 
 
 
 
 
 
 
 CRUSHING STRENGTH AFTER 28 DAYS. 
 
 Fresh water __ 
 
 J i 
 
 8 
 
 287.8 
 
 287.8 
 
 258.0 
 
 215. 5 
 
 ' 137.3 
 
 124.8 
 
 172.8 
 
 From this table we perceive that rich mortars with only 1 or 2 
 parts of sand do not experience any increase of strength through 
 an admixture of trass, even when hardened in sea water (columns 
 1-5). Even the mortar 1 cement to 1 trass is not any stronger 
 than the mortar 1 cement to 1 sand. 
 
 The poor mortar, 1 cement to 4 sand, however, is improved by an 
 admixture of trass (columns 6, 7). 
 
 The tensile strength of mortar of 1 cement, 1 trass, and 4 sand, 
 after 28 days, according to our tests, is not any higher in sea water 
 than in fresh water, while Michaelis finds it to be 9 kilograms 
 stronger in sea water than in fresh water. 
 
 There is, therefore, an important contradiction in the two results. 
 
 A second brand of cement has rendered the same results for the 
 mortar of 1 cement and 4 sand, with and without trass. But in the 
 case of this cement the mortar of 1 cement and 2 sand was improved 
 by an admixture of trass when stored in sea water. 
 
 It is worthy of notice that in the above table the mortars mixed 
 with trass have a comparatively smaller crushing strength, and, 
 again, that an addition of | cement to the mortar 1 : 4 has at least 
 the same effect as one of trass. * * * 
 
 Our preliminary investigations on the employment of trass as an 
 admixture for cement mortar in the case of maritime structures 
 have rendered results essentially different from those of Dr. 
 Michaelis. 
 
 Still we have not arrived at a final opinion in this matter. Tests 
 continued over a long period in the ocean itself will be required 
 to decide whether admixtures of trass are advantageous in cement 
 mortars for maritime purposes. 
 
THE INFLUENCE OF SEA WATER ON MORTARS.* 
 BY E. CANDLOT. 
 
 Since Vicat's important investigations many theories have been 
 advanced concerning the decomposition of mortar by sea water. 
 Although these theories may furnish valuable hints as to the selec- 
 tion and treatment of mortar materials, the results of practice alone 
 form a safe guide for the constructing engineer. 
 
 But practical tests are of value only when they extend over a 
 sufficient length of time. A period of fifteen or twenty years is 
 frequently required to judge of a certain material employed under 
 certain conditions. 
 
 If such practical observations are made systematically and care- 
 fully their value is very great. 
 
 Investigations in the harbor of La Rochelle since 1856 are of 
 great importance in this direction, since they extend over a period 
 of forty years and have been made under conditions adapted closely 
 to practice. We [Candlot] are indebted to Messrs. Thurninger and 
 Viennot, the former chief and the latter ingenieur des ponts et 
 chausse'es, for a summary of the observations made so far, and we 
 believe that the conclusions drawn from them will be read with 
 great interest by all concerned in the question of the influence of 
 sea water on mortars. 
 
 Mr. Viennot intends later on to publish a complete report ren- 
 dering all details and all results found during this long series of 
 investigations. 
 
 A first series of cubical blocks of 60 centimeters length of edge 
 was exposed to the open sea during the period from 1856 to 1875; 
 these blocks were above the water's surface at each low water. The 
 mortars were composed of hydraulic limes of different origin, of 
 natural cements from Ponilly, Vassy, etc., of artificial pozzuolanas 
 mixed with various kinds of lime and sand, of admixtures of lime 
 and cement, and finally of trass from Andernach, mixed with var- 
 ious kinds of lime and sand. 
 
 Nearly all blocks had completely lost their cohesion after periods 
 of various length. The rest were strongly affected or nearly 
 destroyed. Only a few blocks of Portland cement were experi- 
 mented upon, of which the blocks of mortar of English cement 
 (1857), of Portland cement from Dauphine's (1859), and of Portland 
 
 * From TJionindustrie-Zeitung, February, 1897. 
 
70 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 cement from Boulonnais (1875) were in very good condition. Blocks 
 of neat cement (English and French) were decomposed. 
 From these tests Viemiot draws the following conclusions : 
 
 (1) Neat cements are destroyed more rapidly than mortars of a 
 certain composition. 
 
 (2) Mortars made of 1 volume of cement to 1 volume of sand, 
 and again of 1 volume ol cement to 2 of sand (that is, about 1,300 
 kilograms and 650 kilograms respectively of cement to 1 cubic 
 meter of sand) are those which offer the greatest resistance to sea 
 water. They will last for 20, 3G, and 38 years. 
 
 In 1880 Thurninger commenced a new series of tests with blocks 
 of masonry and of concrete made of lime of Theil mortar. The 
 length of edge of the cubical blocks was 40 centimeters. 
 
 On July 1, 1895, the masonry blocks had disappeared, their 
 destruction having commenced four years after their exposure. 
 Out of thirty-two concrete blocks only twenty-six remained, but 
 they were in a state of advancing decomposition and certain sooner 
 or later to crumble away entirely. 
 
 A second series of tests with cubical blocks of 40 centimeters 
 length of edge was also commenced in 1880. These blocks were 
 submerged in sea-connected basins, the water in which, on the aver- 
 age, was renewed twice during each day ; as a rule the blocks do 
 not show above the water's surface. Their number at present is 
 193, and they consist chiefly of various limes, French and foreign 
 cements. Out of nine blocks of natural cement from Lot et- Garonne-, 
 two have crumbled away. Three blocks of cement from Grdnotte 
 were made in 1881, eight in 1882, and six in 1887; of these blocks 
 five have been destroyed. 
 
 Experimenting with slag cements was not commenced until 1891 ; 
 out of twenty blocks submerged between 1891 and 1893, fourteen 
 at present are crumbling away; four others, submerged in 1894 
 and 1895, are still in good condition. 
 
 Out of thirty-one masonry blocks laid in Portland cement mortar 
 and submerged between 1881 and 1892, twenty-two are still intact, 
 while nine have commenced to disintegrate. Among the older 
 blocks there are three laid in English cement mortar (1883) and 
 three laid in Boulonnais cement mortar (1895) of great power of 
 resistance. 
 
 According to Viennot, these tests point to the following con- 
 clusions : 
 
 (1) Mortars of hydraulic lime, mixed in any proportion, in most 
 cases commence to disintegrate after one or two years' immersion 
 in sea water; they crumble into pulp after periods varying in 
 length, but apparently not exceeding fifteen years. 
 
 (2) Concrete resists better than masonry, owing to the greater 
 density imparted to it by ramming. 
 
INFLUENCE OF SEA WATER OX HYDRAULIC MORTARS. 71 
 
 (3) Rapid-setting cements may commence to disintegrate after 
 six or eight years, but again it has been observed that they may 
 last longer than thirty-eight years without crumbling. 
 
 (4) The mortars offering the greatest resistance are those con- 
 sisting of 1 part cement to 1 or 2 parts respectively of sand (1,300 
 kilograms or 650 kilograms respectively of cement per cubic meter 
 of sand). 
 
 This mixture corresponds to the weight of cement required to 
 fill the spaces between the grains of sand. These, therefore, are 
 the least porous mortars. 
 
 A theory of the resistance of mortars to sea water can only be 
 of speculative interest, unless it is based on practical investiga- 
 tions of long duration. The experiments made at La Rochelle, 
 during nearly forty years, form a safe basis of discussion of this 
 matter, and we may now proceed to investigate the question why 
 Portland cement appears to be superior in maritime structures to 
 any other hydraulic binding medium, and, again, what conditions 
 must be filled by mortars in order to make them proof against the 
 destructive influence of sea water. The principal fact clearly and 
 unmistakably proved by the La Rochelle tests is this : That none of 
 the hydraulic binding media so far known will resist the action 
 of sea water if the latter is able to penetrate into the mortar. 
 This conclusion is to be regretted, because it excludes the hope 
 that the question of the behavior of mortars in sea water may be 
 solved within a reasonable time by purely chemical investigations. 
 
 Verification through time tests is always indispensable, as theory 
 is too often contradicted by facts. But these tests have settled one 
 very important point in an unquestionable way, viz, the necessity 
 of employing only such hydraulic binding media as do not contain 
 any free lime, which may slake after gauging. 
 
 Destruction of mortars exposed to the action of sea water may 
 occur in a twofold manner ; either it proceeds from the exterior to 
 the interior, which is commonly the case, or it commences in the 
 interior, in which case it is caused by the hydrating of free lime 
 or free magnesia. 
 
 When the silica and alumina contained in homogeneous cement 
 is not sufficient to saturate the lime* contained in it, then there 
 will be free lime. The total quantity of lime must never exceed 
 the following proportion : 
 
 SioT-SiA <3(H ' leChatelier) - 
 
 Free lime is also found in cements of normal composition, but 
 of imperfect mixture. This will occur in natural cements which 
 
 * Magnesia is always present in small quantities ; the lime only is of importance. 
 
72 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 are mixtures of separately-burnt stones, some containing an excess 
 of lime, others an excess of alumina. 
 
 In hydraulic limes free lime is due to imperfect slaking. The 
 excess of lime is not injurious after it has been slaked. The 
 magnesia, separated from the sulphate of magnesia by the lime 
 contained in the mortar, is also harmless. 
 
 Free lime is most dangerous when contained in compounds burnt 
 at a high temperature. In these cases it absorbs water very slowly, 
 and a considerable time may pass after gauging the mortar before 
 it acts. If there is so much of it that its expansive force is greater 
 than the cohesion of the mortar, then the latter is either torn 
 asunder or cracks are caused through which water can penetrate 
 and complete the work of destruction. If such mortar hardens in 
 air, or in fresh water, the absorption of water may proceed so 
 slowly as to allow the gradually-increasing force of cohesion 
 effectually to resist expansion, provided the percentage of lime is 
 not too great. In sea water, however, the absorption of water 
 proceeds with greater rapidity and the mortar is not allowed to 
 attain a sufficient degree of strength. The expansion caused by 
 free lime in these cases is of a very destructive nature. 
 
 Therefore only binding media containing no free lime should 
 be employed in sea water. In carefully-prepared artificial Port- 
 land cements it is easy to secure this condition. The natural Port- 
 land cements that have been calcined at a high temperature are 
 very dangerous, and their use should be prohibited. Incidents of 
 an exceedingly significant nature, that have occurred during prac- 
 tical use of these cements, can only be explained by expansion 
 caused by free lime. Storage in pits for months, and even years, 
 is not sufficient to secure in every case hydrating of the free 
 lime. 
 
 This has led to ignoring the water's chemical action on mortars 
 and to study their behavior from a physical point of view only. 
 As all mortars, without any exception, are decomposed when per- 
 meated by sea water, the investigation has been restricted to finding 
 the densest mortars, which are the most impervious ones. Labo- 
 ratory experiments, therefore, should be abandoned, as they can not 
 furnish any new information ; accommodation to the conditions of 
 actual use is required to consider simultaneously wave action, 
 difference of temperature, etc. 
 
 Laboratory tests have only rendered one piece of information, 
 which, while being of great importance, could readily have been 
 foreseen ; we refer to the great influence of the quality of sand. 
 The advantages of coarse sand are universally acknowledged to-day 
 and the use of fine sand has been discarded everywhere. 
 
INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 73 
 
 When the density of mortars is studied it will be found that in 
 this quality Portland cement, is very much superior to all other 
 hydraulic binding media. 
 
 We believe the good behavior of the cement mortar gauged 
 in the proportions of 650 and 1,300 kilograms, as has been done 
 at La Rochelle, and for most maritime structures during the last 
 fifty years, is principally due to the density of Portland cement 
 mortar. This binding medium possesses the greatest specific 
 gravity, it requires the smallest quantity of gauging water, and, 
 at the same time, will bind chemically a greater quantity of 
 water than any other cement. 
 
 Dr. Michaelis has recently published an exceedingly interesting 
 paper on decomposition of mortars in sea water. On the basis of 
 the theory that the best resisting cements are those in which lime 
 is found in the most stable compounds, he shows that Roman 
 cements contain the smallest percentage of lime, which is able to 
 act on the sulphate of magnesium present in sea water in forming 
 injurious salts (sulphate and sulphoaluminate of lime). In Port- 
 land cement, during hardening, there separates a certain quantity 
 of hydrate of lime, which strives to change into sulphate or 
 sulphoaluminate of lime. 
 
 It is asserted by Dr. Michaelis that even lime of Theil resists 
 the action of sea water better than Portland cement. We are far 
 from indorsing this theoretical speculation, which is completely 
 contradicted by facts ; the latter demonstrate that Portland cement 
 resists the action of sea water much better than all other hydraulic 
 binding media, notwithstanding the possible inferiority of its 
 chemical composition. 
 
 The importance attaching to the density of mortars has, how- 
 ever not escaped Dr. Michaelis. He says : 
 
 But as Roman cements are burnt at red heat, or at moderate red heat only, 
 during which operation they do not condense, they must be pronounced from a 
 physical point of view to be of a porous nature ; the compounds forming during 
 absorption of water will, therefore, be contained in them in a much swollen 
 state ; hence the mortars produced with them will shrink considerably during 
 air-drying, through loss of loosely-bound water, nearly all water contained in 
 these hydrates, over and above the quantity corresponding to the hydrate of 
 lime being such loosely -bound water. 
 
 Physically the hydraulic limes, the best representative of which is lime of 
 Theil, are quite close to the Roman cements. Their density, at least that of 
 the so-called light ones, is generally less than that of Roman cements. 
 
 From a physical point of view Portland cement is much superior to hydraulic 
 limes, because it acquires great density through vitrification at white heat. 
 During set the pores are filled more completely because, the particles being closer 
 together, there is in the same space much more swelling substance. The mean 
 proportion of mass in equal spaces is for Portland and Roman cement about 5 : 3, 
 and for Portland cement and hydraulic limes from 5: 2.5 to 5: 2; vitrified Port- 
 land cement, therefore, has a much greater volume weight, and consequently its 
 
74 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 mortar attains a much higher degree of density, or rather of condensation, since 
 Roman cements and hydraulic limes may likewise present a perfectly close grain. 
 
 And again Dr. Michaelis adds : 
 
 It is evident that closeness of grain in other words, imperviousness is a very 
 important consideration. 
 
 Even if the quantity of binding medium employed completely 
 fills all voids in the sand, there will never result a mortar of perfect 
 density. There is always a certain quantity of voids. In gauging, 
 air bubbles remain inclosed in the mass and are not removed even 
 when the greatest care is exercised. The volume of voids in bind- 
 ing media is about the same in all of them. The considerable dif- 
 ferences in the porosity of mortars are caused by voids resulting 
 from the excess of gauging water. In the first place the various 
 binding media require very different percentages of water for gau- 
 ging, and in the second place the volume permanently absorbed as 
 water of crystallization differs very much. 
 
 To obtain a paste of standard consistency, there are required for 
 Portland cement 25 per cent of water and for limes and Roman 
 cements from 40 to 60 per cent of water. Under ordinary circum- 
 stances there are absorbed as water of crystallization by Portland 
 cements from 18 to 20 per cent, by Roman cements and hydraulic 
 limes from 8 to 10 per cent, and by slag cements from 5 to 6 per 
 cent. We perceive from this that the voids caused by excess of 
 gauging water in Portland cement amount to little or nothing, 
 while in other hydraulic binding media they will run up as high 
 as from 30 to 40 per cent. In mortars, the sand which is added to 
 the cement will reduce the advantage of Portland cement in this 
 respect. For fine-grained sand a great deal of gauging water is re- 
 quired, whatever may be the nature of the binding medium. When 
 the sand is coarse and the percentage of cement high, then the 
 amount of water required for gauging is not much more than that 
 required for neat cement.* 
 
 It has been shown by Alexandre that in lime mortars the volume 
 of voids, that is to say, the difference between the apparent and 
 the actual volume of mortars after set, varies between 23 and 31 
 per cent; in cement mortars it varies between 13 and 31 per cent, 
 the former figure corresponding to a mixture of 550 kilograms of 
 cement with 1 cubic meter of coarse sand, and the latter figure 
 corresponding to a mixture of 250 kilograms of cement with 1 cubic 
 meter of fine sand. 
 
 Mortar made of 650 kilograms of cement and one cubic meter of 
 coarse sand was found to contain only 9 per cent of voids ; for lime 
 
 * Very explicit notes on this subject will be found in the writings of Alexandre 
 (Annales des ponts et chaussees, September, 1890) and Feret (Annales des 
 ponts et chaussees, Juillet, 1892). 
 
INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 75 
 
 the same percentages rendered 24 per cent of voids ; it may be added 
 that the density of lime mortars is not influenced much by the pro- 
 portions of the mixture ; indeed, the rich mortars generally are more 
 porous than those of an average degree of strength. In cement 
 mortars, however, density increases in proportion to the percent- 
 age of cement, and, in addition to the advantages of density, those 
 of strength are attained. 
 
 If rammed concrete is employed the voids may be still further 
 reduced. The advantages of concrete as compared with masonry 
 are also shown by the experiments made at La Rochelle. As its 
 price has not increased appreciably, it is easy to understand that 
 its use has become more and more universal, especially in foreign 
 countries, where all great maritime structures have been carried 
 out in cement concrete. 
 
 It seems surprising that blocks of neat cement should have been 
 destroyed, while those of mixtures of 1:2 have stood well. Mor- 
 tar made of neat cement, it is true, is of great density and in a 
 short time becomes quite impervious. But all those who have 
 made any tests of strength with such mortars, after they had been 
 exposed to sea water, will know how fragile they become after a 
 few months. Although the briquettes do not show any outward 
 sign of destruction, they have become as brittle as glass and crack 
 under the slightest pressure ; to this phenomenon, which doubtless 
 must be attributed to excessive crystallization, is due the distrac- 
 tion of neat cement. When cement contains an excess of lime 
 (which is especially the case when gauged neat) the phenomena of 
 expansion make their appearance with greater intensity, and de- 
 struction is rapid even in calm water. As soon as cement is mixed 
 with sand this fragility disappears ; after four years' treatment in 
 sea water, for instance, we found neat cement to have only 6.4 
 kilograms crushing strength per square centimeter, while the same 
 brand of cement, mixed with sand in the proportion of 1:1, was 
 found to have 54.7 kilograms crushing strength. 
 
 Dr. Michaelis, whose paper we have mentioned above, believes 
 that the solution of the sea- water problem consists in mixing a 
 certain percentage of trass or pozzuolana with Portland cement. 
 The pozzuolana will combine with the free lime, will contribute 
 towards hardening, and will prevent the lime from forming 
 dangerous compounds under the influence of the sulphate of mag- 
 nesia contained in sea water. 
 
 According to Michaelis, no other mortar excels a mixture of 
 Portland cement and trass in regard to price, power of resistance, 
 binding qualities, and good behavior in sea water. In proof of 
 this opinion, he has made a long series of experiments, on the 
 basis of which he concludes that an admixture of trass to cement 
 
76 INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 
 
 mortar increases its strength, and also its resistance to decompo- 
 sition. It is highly probable that pozzuolaiias will combine with 
 the lime contained in cement, and the interest attached to the ex- 
 periments of Dr. Michaelis can not be denied. But, without enter- 
 ing into a detailed discussion of his conclusions, we will demon- 
 strate how difficult it is to interpret such laboratory tests. 
 
 In the above-mentioned paper of Alexandre there are contained 
 a number of comparative tests of mortars gauged with different 
 kinds of sand and immersed in sea water. Now, the sands which 
 rendered a very considerable increase of strength and which pre- 
 vented or delayed decomposition of mortars, were lime sands. 
 Here no chemical influence of sand can be considered, as carbonate 
 of lime could not act on the lime becoming free during gauging and 
 hardening. These results have been confirmed by Feret's tests ; 
 the tensile strength of mortars gauged with the same brand of 
 cement, in the proportion of 1:3, after one year's immersion in sea 
 water, was found to be 25.7 kilograms per square centimeter where 
 standard sand had been used, 39.5 kilograms per square centimeter 
 where crushed marble had been used, and 34.5 where trass had 
 been used. Marble has therefore rendered more strength than 
 trass ; yet there is in this case no chemical action. 
 
 Finally, we have obtained the following results with quartz 
 sand, about whose chemical nonactivity there can certainly be no 
 doubt : Standard sand mortar 1 : 3, after three months' hardening 
 in sea water, 26.5 kilograms per square centimeter; standard sand 
 with 10 per cent fine sand, 32.8 kilograms per square centimeter; 
 standard sand with 20 per cent fine sand, 40.4 kilograms per square 
 centimeter. 
 
 If exceedingly fine powder of any kind whatever was added to 
 the sand, we have invariably noticed an increase of strength, due 
 solely to greater density. 
 
 All these tests show, therefore, conclusively that in all admix- 
 tures to cement, improvement may be due to purely physical causes, 
 and there is nothing to show that admixtures which can combine 
 chemically are advantageous. Conclusions based on these labora- 
 tory tests are therefore of a very dubious nature ; there are too 
 many contradictory points not yet cleared up, and time frequently 
 changes even those opinions which were considered to be the best 
 founded. 
 
 Is the lime becoming free during gauging and hardening inju- 
 rious? Is it advisable to mix a substance with the cement which 
 will combine with this lime ? Do the pozzuolanas exercise a bene- 
 ficial influence? It can not be said that any actual proof has been 
 rendered. 
 
INFLUENCE OF SEA WATER ON HYDRAULIC MORTARS. 77 
 
 The faultless behavior of Portland cement mortars which have 
 been exposed to sea water for more than forty years shows that 
 they will resist decomposition without any admixture of pozzuo- 
 lanas. One fact has certainly been demonstrated beyond a doubt, 
 viz, that mortars should be as impervious as possible. This leads 
 to very simple conclusions, the most important question now be- 
 coming that of the selection of sand. 
 
 The price of Portland cement has depreciated very much, and no 
 excessive expenditure is required even for mixtures of 600 and 650 
 kilograms of cement to 1 cubic meter of sand, mixtures which appear 
 to be necessary. In consideration of the great strength of these 
 mortars, great savings are practicable by reducing the thickness of 
 
 walls. 
 
 * * * * * * * 
 
 Unfortunately, good sand is not found everywhere, and to avoid 
 the cost of transportation, it is frequently necessary to employ the 
 sand found on the ground. 
 
 In his paper on the density of mortars, Feret says : 
 
 We arrive at the conclusion that in order to create the most favorable condi- 
 tions for the resistance of mortars to the destructive action of sea water, the use 
 of sand containing many fine grains has to be avoided as much as practicable, 
 and in case no other sand is available, the quantity of cement used for gauging 
 should be increased. 
 
 We believe that the sand question should be still further explored. 
 Fine sand must decidedly be discarded, and only very good sands of 
 known origin should be used. 
 
 ******* 
 
 We are convinced that rich mortars gauged -with coarse sand, 
 and the exclusive use of concrete, will produce masonry of great 
 power of resistance, the durability of which may be taken as war- 
 ranted. 
 
 6" ', :::V*..".-.VJ A 
 
THIS BOOK IS DUE ON THE LAST DATE 
 STAMPED BELOW 
 
 AN INITIAL FINE OF 
 
371950 
 
 UNIVERSITY OF CALIFORNIA LIBRARY