It must be understood that the above rattling tests are only comparable with each other, group by group, and not group with group. As the charges consisted only of the whole bricks left from the previous rattling of each charge, and therefore varied from 6 to 9 bricks per charge, it will be impossible to accurately connect the performance of each charge with its rate of loss before freezing. It is only possible to compare it with another charge of the same absorption, same number of bricks and worn to about the same degree by an identical treatment, except as to freezing.

A scrutiny of the results shown in Table V develops the following points:

1st. After three minutes' rattling, which should be especially destructive to bricks whose surface had been weakened by repeated freezing, the nine charges stood as follows:

Cases where frozen losses exceeded unfrozen.... . . 4—K, J, F, E. Cases where losses were about equal.....

2-M, G.

Cases where unfrozen losses exceeded frozen...... 3—L, I, H.

The our frozen samples which lost more heavily than their unfrozen companion charges, varied from 8.21 per cent. absorption to 0.77 per cent. The three cases where frozen charges stood more than the unfrozen, had absorptions of 10.19, 5.07 and 4.10 respectively. The charges of equal loss had absorptions of 12.84 and 3.20. In no case did the loss of one charge exceed that of its compnaion charge by a large amount, or one beyond the usual limit of fluctuation of unfrozen charges.

Nothing can be shown from this data, except that the freezing has not affected the softer bricks more than it has the harder ones. 2d. After six minutes' rattling, during which the loose material, if any existed, should have been pretty well worn off from the exterior of the frozen brick, the records show: Cases where frozen losses exceed unfrozen Cases where losses are about equal:... Cases where unfrozen losses are highest

4-K, J, F, E. -3-M, L, I. 2 -H, G.

It will be noted that there was a shifting around of positions during this test, G dropping down to the heavier losses, while L and I become equal.

3d. After ten minutes' rattling, beyond which it was thought not worth while to go, the charges stood as follows: Cases where frozen losses exceeded unfrozen ...2-J, E. Cases where losses are about equal .....

.4—M, K, G, F. Cases where unfrozen losses exceed frozen .....3-L, I, H.

A still further shifting has here occurred, indicating the tendency of the charges to become more nearly uniform in rate of

wear.

These facts seem to show that in this instance and under these conditions that the freezing treatment given was insufficient to produce any clearly marked effect. The charges which lost most were not grouped systematically towards either high or low absorption-in fact, their erratic distribution all over the range forces one to believe that the inherent irregularities of the material were more influential than the effect of frost, and that the variation observed resul.ed from this inherent irregularity and do not indicate anything else.

This frost treatment is unsatisfactory to the writer because its results do not correspond with the general observation that soft burned clay wares do deteriorate on freezing. It was expected

that charges M, L and probably K would show marked deterioration, but it was also confidently expected that charges I, H, G, etc., would not. The results were somewhat disappointing in that no clear cut distinction can be drawn between the hard and soft end of the series.

There remains two courses to pursue: 1st. To devise a much more sudden and heavy freezing test, by which bricks previously soaked full of water can be rapidly reduced to a point well below freezing. 2d. To expose the material to nature's alternations of freezing and thawing. In the case of this investigation the latter course is available, and the charges will half of them be exposed next winter, and in the spring be re-rattled in comparison with the protected charges, to see if any further clue can be gotten. This method, unfortunately, is too slow for a laboratory process and gives no relief in the matter of using the freezing test for determining practical questions when they arrive.

SUMMARY.

This investigation, in the opinion of the writer, shows the following points:

Ist. That this clay, under the manufacturing treatment given by this company, reaches its maturity as regards toughness and ability to resist impact and abrasion, at an absorption of about 5 per cent.

2d. That at this stage, its color is still light red, and its fracture far from vitreous, and it would be condemned by most engineers and most paving brick manufacturers as "too soft."

3d. That no important or consistent gains in strength or toughness are found on burning the material harder, through the dark red, and chocolate-colored stages, where the absorption falls below 1 per cent.; but neither is there any marked falling off. The gains in density and hardness offset the losses by increased brittleness, so that rattling losses remain about uniform during this period.

4th. Burning beyond the chocolate colored stage is invariably a source of deterioration. The hardness and density reach their maximum, while the clay is still dark red. Brittleness continues to increase, and with it a vesicular structure begins to appear, so

that the brick fails both by breaking, chipping and wearing. The percentage losses on rattling do not increase as rapidly from overburning as from underfiring, but the percentage of broken bricks increases very much more rapidly, and they are a very serious cause of street failures.

5th. The weather resisting power of those bricks which have reached their strength maturity, while still of relatively high absorption, has not yet been conclusively proven. Five consecutive heavy freezings and thawings failed to weaken their structure. Whether a still larger number of repetitions would do so is not yet known. The test would be accepted with much more confidence if other still softer samples, which have not reached their strength maturity, had been damaged by these same freezing tests. Such was not the case, however, and this must mean, either that these freezing tests were inadequate, or that the material becomes frostproof a long time before it reaches its best strength, and at a much greater absorption percentage than has been considered possible hitherto.

As a whole, in the opinion of the writer, the value of the rattler test as a means of judging the value of paving material has been clearly proven, and the contention that bricks which will endure the standard rattler test creditably will also endure the effect of frost creditably has been materially strengthened, but not yet conclusively proven.

In conclusion, the writer desires to give public thanks to Mr. Lester Ogden, who has carried out the large amount of work in this investigation with the greatest fidelity and attention to detail.

NORMAL CONSISTENCY TESTS OF NEAT CEMENT.

BY RUSSELL S. GREENMAN.

It is not because I expect to give you any information of value that I am about to open a discussion on this subject of "Normal Consistency Tests of Neat Cement," but rather because of a desire to obtain information for the department with which I am connected. The line of discussion which I shall undertake will be the relationship of normal consistency of neat cement to the time of setting.

Our laboratory being an unknown quantity to most of you, I hope you will pardon a brief statement of our interest in this subject because of the conditions as we find them. Perhaps such a statement will enable you to better understand our desires in regard to the information sought.

As you undoubtedly know, the State of New York is just com mencing the construction of a barge canal, estimated to cost $101,000,000; the people of the state are about to vote upon a plan of issuing $50,000,000 worth of bonds for good roads-with the chances in favor of carrying the same; and the state is constructing a large number of buildings, such as armories, hospitals and schools. For all these projects the laboratory I represent tests the cement. Roughly we figure that during the next ten years, the barge canal will use an average of 350,000 barrels of cement per year, and the good roads will yearly average 140,000 barrels. The public buildings, constructed under the direction of the State Architect, will average about 25,000 barrels of cement each year. By testing a sample from every tenth barrel we will have to test over 50,000 samples each year.

In the case of extensive work by a corporation it is possible to specify a particular brand or brands. In the case of public work, as you know, the state cannot, or rather should not, specify brands so during the past five years we have tested 58 brands of cement; and each yearly average is about 32 brands.

The various contracts call for lots of cement ranging from 125,000 barrels to 100 barrels, but it is safe to say that the ordinary contractor on good-road work will buy his cement only as he has