and D. It is due to the north side of the canal being much rougher than the south side, the concrete of the north side having been cut away preparatory to its repair just previous to making the experiments of series C.

FIG. 21.-Curves of equal velocity in experiment No. 14, series C. The dots are for small Price meter No. 351; heavy black squares for small Price meter No. 363.

Vertical velocity curves for these four series of experiments (A to D) are shown in figs. 23 to 28. Each of these curves is the mean of observations in the eight verticals 1, 3, 5, 7, 9, 11, 13, and 15 feet from the south side of the canal. For series C and D, figs. 26, 27, and 28, the observations obtained with small Price meter No. 363 at six-tenths

FIG. 2.-Curves of equal velocity in experiment No. 20, series D. The dots are for Haskell meter No. 3; heavy black squares for small Price meter No. 363.

depth are shown by the little squares. As a rule they fall outside the curve, meter No. 363 indicating a somewhat greater velocity than meter No. 351. This fact is also shown by the corresponding curves of figs. 16 and 17. The two sets of velocity curves (full and dotted) of

Two

fig. 28 were found from simultaneous velocity measurements. parties measured velocity in the same section at the same time. One party, with the Haskell meter, started near the south side of the canal and worked toward the north side; the other party, with the Price meter, started at the north side and worked toward the south side. It is seen from these curves (fig. 28) that the bottom velocities agree more closely than those at the surface, and that the Haskell meter

FIG. 23.-Vertical velocity curves obtained with current meters in experiments in series A. Nos. 1, 2, 8, 12, and 13 are for large Price meter No. 88; Nos. 4, 5, 6, 7, and 11 are for small Price meter No. 351.

gives a much larger surface velocity than the Price meter. We should expect a little difference, since the revolving head of the Haskell meter is larger than the wheel of the Price meter, and its center being farther below the surface it must revolve faster. But the difference in velocity is much greater than this little difference in depth of axis would indicate. (See p. 91.)

Fi. 21.-Vertical velocity curves obtained with Fteley current meter in experiments in series A.

The shape of these curves evidently depends on the depth and velocity. For depths not exceeding 1 foot it is a nearly straight line, with the maximum velocity at the surface and the mean velocity at mid depth. For depths of 8 or 9 feet and small velocities the curve is very flat, with the maximum velocity at two-tenths to three-tenths depth and the mean velocity at sixty-five-hundredths to seven-tenths depth below the surface. The radius of curvature decreases as the velocity increases. In all of these curves the least velocity is at the

bottom. Nos. 6 and 8, series B, fig. 25, show the effect on the vertical velocity curve of checking the discharge at the lower end of the canal. For all the other curves of fig. 25 there is free discharge. The effect of checking the discharge at the lower end is to force the thread of maximum velocity from the surface to a depth of two-tenths or more. The table on page 77 gives the position of the threads of maximum and

FIG. 25.-Vertical velocity curves obtained with current meters in experiments in series B. Nos. 1, 15, 17, and 19 are for small Price meter No. 351; Nos. 3, 5, 7, 8, 12, and 14 are for large Price meter No. 88; Nos. 2, 4, 6, 9, 10, 11, 13, 16, 18, and 20 are for Fteley meter.

mean velocity taken from the curves shown in figs. 23 to 28. It is difficult to locate the threads with accuracy, since very little change in the shape of the curve would make a great change in the position of the threads. It is easily seen that the shape may be changed somewhat, and yet fit the observations nearly as well as the one drawn and give a discharge differing from it but little. For velocities of a foot

FIG. 26.-Vertical velocity curves obtained with current meters in experiments in series C. Heavy round dots are for small Price meter No. 351; the square blocks are for small Price meter No. 363.

or more per second, however, the threads can be located with a fair degree of accuracy. The position of the thread of maximum velocity as found from series A is eighteen-hundredths depth below the surface; for series B it is one-tenth depth; for series C, thirty-eight-hundredths depth; and for series D, twenty-four-hundredths depth. For the mean

of C and D, in which the conditions are nearly the same, it is thirtyone-hundredths depth. The position of the thread of mean velocity as shown by this table is sixty-six-hundredths depth for series A, fifty-four-hundredths depth for series B, and sixty-six-hundredths

FIG. 27.-Vertical velocity curves obtained with current meters in experiments in series C. Heavy round dots are for small Price meter No. 351; the square blocks are for small Price meter No. 363.

depth for series C and D. It is seen that the threads of mean and maximum velocity are nearer the surface in series A than in series Cor D, although the depths do not differ much. This is probably due to the influence of the Croton model dam, the long upstream slope of

FIG. 28.-Vertical velocity curves obtained with Haskell meter No. 3 and small Price meter No. 363 in experiments in series D. Open circles are for small Price meter No. 363, point method; heavy black dots are for Haskell meter No. 3, point method; heavy black squares are for small Price meter No. 363, six-tenths-depth method.

which has had the effect of forcing the thread of maximum velocity up to thirteen-hundredths depth. As the ratio of width to depth changed from 16 to 2 the thread of mean velocity moved from fifty-fourhundredths depth to sixty-six-hundredths depth below the surface.