tained upon the jack gages when the pumps were operated at normal speed.

In some respects the manner of cracking of all these pipes was very similar. Usually cracks appeared in either or both top and bottom before they did in sides. In no case did cracks appear in top, bottom and sides, and in the bell end at the same time. In three cases, Nos. 708, 709 and 713, cracks appeared in top, bottom and sides, in the spigot end, at the same time. In one case, pipe No. 712, cracks appeared in top, bottom and one side in the spigot end at the same time. These facts indicate that the pipes were weaker at the spigot end.

It was interesting to note that in the bar reinforced pipes cracking, especially on the sides, was confined to a few main cracks (See Fig. 3), while in the mesh reinforced pipes the cracking in the sides was much more distributed and irregular. (See Fig. 4.) In all cases the cracking in top and bottom was confined to a less number of cracks than in the sides. In those pipes in which the concrete appeared denser and of better quality the cracking did not start at as low a load, but generally developed more rapidly. There seems to be no fixed relation between the load at which cracking begins and the maximum load.

All difference in the character and number of cracks seems to be traceable to the amount, kind and location of the reinforcing. It has been stated that cracks usually appeared in top and bottom before they did in the sides. This is to be expected when such pipes as these are tested with the distributed sand bearings, as the moments were greater at the top and bottom than at the sides and the reinforcing is less effective. When the pipes are tested the top and bottom portions, also the sides, tend to flatten out, causing a tendency for the reinforcing to pull out of the pipe wall. At the top and bottom this pulling out tendency is resisted by a comparatively thin layer of concrete (See Fig. 5). The reinforcing in the sides is subjected to the same tendencies but here the pulling out is resisted by a thicker layer of concrete.

The main failure cracks at the sides occurred at some

distance above or below the center line and quite close to the point where the reinforcing crosses the neutral axis of the pipe wall. The point of zero bending moment for a circular ring loaded as in these tests is near the 45° point. In these pipes the reinforcing crosses the neutral axis of the pipe wall at 212, 521⁄2 and 81⁄2 inches below and above the 45° point for the 18 inch, 30 inch and 48 inch pipe, respectively, or 5, 9 and 1411⁄2 inches above (and below) this point. (See Fig. 5).

One pipe was reinforced with corrugated bars, but there was nothing to indicate that this reinforcing was any more effective than the smooth bars. When bell pipes are laid with uncemented joints, a portion of the spigot end of

Fig. 4. Pipe No. 705 after the Maximum Load, Showing the Typical Manner of Crushing of Mesh Reinforced Pipe. Note the Location of the Principal Side Failure with Reference to the Center-Line.

each pipe receives no load. When these same pipes are laid with cemented joints, each end of each pipe is equally rigid and will deflect the same amount. In an effort to approximate both the above conditions and at the same time to adhere to the standard specifications for loading, which requires a full length loading, the center of the load was placed at a small distance from the center of the pipe toward the bell end. In the 48 inch pipe a loading 41/2 inches off center seemed to give nearly equal deflection. However, these tests were not extensive enough to determine either the effect or the proper distance of off center loading.

By observation of the pipe under test and by a study of the collected data it was clearly seen that at some load less than the maximum load, a loading was reached beyond which the pipe would be unsafe. There was a decided change in the direction of the load-deflection curves at a load near and usually below half the maximum load. This load, taken from the load-deflection diagrams, was found to come just before the load at which the first full length crack appeared, except in two cases, where it came just after the load that produced the first full length crack. This load was termed the "Critical Load."

Standard absorption tests were made upon two pieces taken from each tile tested. There seems to be no fixed relation between the absorption and either the critical or maximum loads.

The records of the deflections of these pipes showed that at the critical load the elongation of the horizontal diameter was from .01 inch to .05 inch and at the highest load at which it was possible to measure the deflection it was from .04 inch to 1.2 inches, with an average of from

Fig. 5. Pipe No. 702 after the Maximum Load, Showing the Pulling Out of the Reinforcing and the Crushing at the Compression Faces on the Sides. The Crushing was Never Noted Until, or After, the Maximum Load.

.05 inch to .75 inch. It is readily seen that for lateral extension of .01 inch to .05 inch the ditch filling would not be sufficiently compressed to give any lateral support. This is especially significant when considered in connection with the fact that the critical load was, approximately, the load at which a full length crack was developed.

As far as this series of tests are concerned, certain points seem to be brought out quite clearly. The pulling out of the reinforcing was usually noticed when the pipes were subjected to a load higher than the critical load, but it seems reasonable to suppose that the bond between the concrete and the steel had been impaired before the pulling out occurred, thus bringing these two loads nearer together. It seems evident that some method of anchoring the reinforcing at the top and bottom would have made the steel more effective. The development of the principal side cracks so near to the point where the reinforcing crosses the neutral axis indicates that the reinforcing would have been more efficient had it been so located as to cross the neutral axis of the pipe wall at the 45° point. Although no definite turning point was evident during the tests, the data show that the safe load for these pipes was somewhat less than onehalf the maximum load.

However, this series of tests is so limited that no general conclusions can be drawn. The above statements are applicable to the specimens tested, but should not be considered as applying to reinforced pipe in general.

An Investigation of Causes of Breaks in Large Water Mains in the City of Chicago

Editor's Note.-The following material is an abstract from the engineer's report in the general report of the Department of Public Works of the city of Chicago for the year 1913, which report is just published. Records of serious breaks have been made each year, but no other extended examination of the causes was ever made before. Following a series of disastrous breaks in the spring and summer of 1912, the examination was ordered with a view of discovering the cause or causes, and suggesting remedies and preventive measures.

Explanations of these causes follow. After the explanation is a discussion of the breaks with a view to determine the possible causes which are the most probable ones in the cases under study.

EXPLANATION OF CAUSES.

1. IMPROPER DESIGN.

For the design of pipes there are numerous formulas giving very nearly the same results. Pipes have been manufactured in such quantities that these formulas are as