his constructions. He did once use it for some below-ground work on a bridge in Southern California, in which every legitimate effort had to be made to keep the total cost of structure within a limit of $200,000. There proved to be a margin of less than $500 when the bridge was finally completed and turned over to the owners.

Concretes as poor as 1 : 4 : 8 or even 1 : 6 : 10 have been used in times past in large mass-work, in order to reduce the cost of construction; but, in the author's opinion, such mixtures are not legitimate. Some tests on plain concrete beams by Wm. B. Fuller, Esq., gave on the average the moduli of rupture indicated in the following table:

Table 39b

This table gives a very clear idea of the relative strengths of concrete of varying richnesses; and it is evident therefrom that the last two mixtures are too poor to warrant use in any first-class construction, also that 1 : 2 1/2 : 5 concrete is better than 1 : 3 1/2 : 5 concrete in the ratio of 1.33. By interpolation it may be concluded that it is probably better than 1 : 3 : 5 concrete in the ratio of about 1.2. Comparing the 1 : 22 : 5 mixture with the 1 : 3 : 6 one, the ratio of strengths would be about 1.5.

Reduction of Voids in the Aggregate

Whenever there is a large amount of concreting to be done on a job, it will prove to be economical to study carefully the percentage of voids in the aggregate of broken stone and sand (or of gravel and sand), and to experiment in order to determine what mixture of broken stone, and gravel, or of several sizes of broken stone, with the proper amount of sand in each case, will reduce the said percentage of voids to a minimum. If then this mixed aggregate be adopted, and if the amount of cement is never less than one and a tenth times the volume of voids, the resulting product will be first- class and satisfactory to all concerned, provided, of course, that only proper materials be employed and that the mixing be thorough.

As a matter of precaution, however, against carelessness or error on the part of the tester of voids, the author's specifications require that, for aggregates in which all the materials are measured separately before mixing,