Mr. Launcelot Andrews, Ph.D., in an article on cements in Clay Record says that an ideal Portland cement should be composed of:

Lime
Silica

Alumina

Per cent.

62.2

28.2

9.6

but adds that about a third of the alumina may be replaced by ferric oxide, which would correspond to the composition:

He also gives 3 per cent of magnesia as the maximum to be allowed, a larger amount having a tendency to cause the cement to swell and crumble.

Table No. 15 shows the composition of several well-known American cements, also taken from Cumming's "American Cements."

It will be noticed that the two brands of Louisville very quicksetting cements are high in lime and correspondingly low in magnesia, that there is a difference between the naturals and Portlands in every essential ingredient, and that it is so marked that the one can always be distinguished from the other.

Fineness.

Besides its composition, there is another property of cement which has an important bearing upon its value in mortar, and that is its fineness. It costs materially more to grind a cement so that 75 per cent of it will pass a sieve of 40,000 meshes per square inch than to pass one of 10,000, so that the tendency is to leave the product as coarse as possible and get satisfactory results. Gillmore says: "The capacity of a cement to receive sand, other things being equal, varies directly with its degree of fineness." As cements are always used in practice mixed with a certain amount of sand, this matter is of great importance. The author just quoted says that not more than 8 per cent of a cement should be rejected by a sieve of 6400 meshes to the square inch. Mr. Andrews, previously referred to, says that all grains so large as not to pass a sieve of 75 meshes to the linear inch (5625 per square inch) should be considered as inert or wholly passive constituents, and

that they should not constitute more than 20 per cent of the total weight.

Mr. R. W. Lesley in examining different specifications upon. this point found the requirements as shown in Table No. 16 (the results being given in a paper read before the Engineers' Club of Philadelphia).

In prosecuting the Boston Main Drainage Works, Mr. Eliot C. Clarke made some very elaborate experiments to show the effect of fine grinding on cements. In Tables Nos. 17 and 18 are given some of his results. The figures represent the tensile strength in pounds per square inch.

In Table No. 18 the same brand was used in both cases, but one sample was taken from the ordinary delivery, and the other from a lot that had been ground in accordance with a special contract.

Another test was made by taking the average of these brands of finely ground with the same number more coarsely ground, with the results shown in Table No. 19.

These tables show conclusively the value of fine grinding, and, as far as investigations have been carried, that the finer the cement.

is ground the more strength it will have when mixed with sand. On account of the great cost of extreme grinding, it is not economical to carry it too far. From the figures previously given, it would seem that the authorities had decided upon a sieve of 200 meshes to the linear inch as the limit to be required.

Concerning the tests to be made of cements to determine its real value or its special fitness for any particular work, there is much to be said. Different engineers have different requirements when seeking for the same results, and different laboratories differ very much among themselves in their methods, and consequently their results vary materially even when cement from the same barrel is used. The best illustration of this is shown in Table

No. 20, taken from a paper by Prof. J. M. Porter of Lafayette College. Prof. Porter had ten samples taken from the same number of barrels of Portland cement, thoroughly mixed, and then divided into ten smaller portions which were sent to ten different persons with a request that a seven-day tensile test, one cement to three sand, be made according to the standard of the American Society of Civil Engineers.

These results would seem to indicate that such tests are of little value when a report from one laboratory would cause the cement to be rejected without hesitation under ordinary specifications, and as unhesitatingly accepted according to the report of another equally reliable, and when a special effort has been made to have all conditions as nearly alike as possible. This is hard to explain. But on account of these variations tests of cement must not be given up, but continued with more care, and perhaps on different lines.

It is rarely possible to give the cement used in any large and important work sufficient tests to demonstrate its absolute fitness. It must be done analogically. It is necessary, however, to find a brand of cement before the work is begun that either by experience or long-time tests has been proved to be all that is required. If the former, a series of tests should be made extending over a sufficient