15. No. 4, after a few days' drainage, would doubtless have had its voids reduced to 53 or 52 per cent, which would have raised its weight to 70.1 or 71.6. Likewise No. 7, after a few more days' exposure, would have voids of probably 54 per cent and weight of 73.5 pounds. No. 5 can represent but a small fraction of surface soil deposited by the unrestrained river, for it is to a degree a mixture of finer silt with the coarser and heavier canal dregs; yet it would represent a larger fraction of deposits within a restrained river unable to grade its earthy material.

16. Nos. 1, 2, and 3, from the Picacho Reservoir, do not contain the heaviest material, for that had lodged in the feeding canal or had passed over the diversion dam. Nor do they contain a full quota of the finest and lightest material which reached the reservoir for, during portions of the years the reservoir has been in service, water simply passed through, leaving its heavier silt in the trough where the samples were taken and carrying the rest to make somewhat lighter soil at Casa Grande.

17. If in No. 9 the percentage of voids, made large by the activities of organic life, were restored to 55 per cent, the weight would rise to 76.7 pounds, which is not much lighter than No. 11, which is 15 feet. under No. 9. Were it warranted to theorize on so small a difference as a pound it could be suggested that No. 10, being in the bed of the wash, had opportunity to swell on being wetted after relief from load, while No. 11 was dug for, 2 feet into the side wall at its foot, and the few grass roots, together with swelling, would account for the lighter weight of No. 12.

18. Nos. 9, 10, 11, and 12, being all near the river, which on account of the hard bluff on the opposite bank could never have been far away, are the best located for the depositing of material of high specific gravity, and for the most complete grading of particles, resulting in high percentage of voids. No. 13 is heavier than No. 14, because it was dropped on a steeper grade by waters from a tributary. No. 15 represents but a small part of what the Gila River would deposit behind a dam, yet it is quite possible that such sand would amount to one-fifteenth of the whole. Its voids approximate the 43 per cent found in many carloads of run-of-crusher stone, after being jarred by hauling a few miles on a rough track.

19. Assuming that all silt deposits will have on occasional years opportunities to shed excess water, and that only an insignificant part of the whole will be afterwards made lighter by the activities of organic life, it appears from the foregoing table and discussion that the Gila River, without a dam, would drop silt at San Carlos, averaging at least 76 pounds of dried material to the cubic foot of comparatively shallow deposit, and that it would convey by way of the Florence Canal silt making surface soil which would average 70 pounds. 20. The average of the specific gravities of soils in the above table. is about 2.54. Averaging the specific gravities of silt passing the dam site, as recorded further along in this paper, gives about 2.46. The difference is due to the lightest material having been carried to a more distant field.

21. If as a result of the impounding dam the heavy and coarse. silts were thoroughly intermixed with the fine, the percentage of voids, even in surface soils, would approach that in ordinary building sand, or about 43 per cent, which in connection with a specific

gravity of 2.46, would make soil of 87 pounds. But there would not be such thorough intermixing, except perhaps in a portion of the reservoir, for while the reservoir is yet large, only the finer grades of silt would reach the lower portions of its area, especially its margins, the coarser material being deposited mostly at its head. But it is probable that there would be enough intermixing to raise the weight of surface soil to an average of 80 pounds, corresponding to 48 per cent voids.

22. There has not been an opportunity to obtain within reasonable cost the weight of river deposits at great depths, but it is not unreasonable to think that at depths of 100 feet the weight would be at least an eighth more, or 90 pounds, for it seems probable that a 9-inch layer of surface soil would be compressed to 8 inches or less if loaded for years with a column of earth 100 feet high. Equally reasonable is it then that old deep deposits filling a large reservoir here would average 85 pounds, which is the figure sometimes used by engineers.

23. Though the 85 pounds would now be used with much confidence in computing the life of the proposed San Carlos Reservoir, conservatism requires the use of as low as 70 pounds in calculating the amount of annual dredging that would be required to maintain the reservoir capacity; for the greater portion of the dredging would be in the less compressed comparatively shallow deposits, and a part of it would be of still lighter mass from which excess water had not had an opportunity to escape.

24. To illustrate the effects of diameter and length of settling tubes on the indicated volumetric percentage of mud, several experiments were made in which various tubes were filled from the same sample of muddy river water, and read daily. Three of the results are tabulated, as follows:

TABLE II.

Indicated volumetric percentage of mud after settling.

1 day. 2 days. 3 days. 4 days. 5 days. 6 days. 7 days. 8 days. 9 days. 10 days.

25. The mud from Nos. 7 and 8 of September 6, when dried, had a specific gravity of 2.47, and was by weight 1.52 per cent of the water from which it was deposited, and would produce 1.35 per cent of soil in place, of which a cubic foot after being dried weighs 70 pounds, or 1.12 per cent of 85-pound soil. The mud from Nos. 7 and 8 of September 9, when dried, had a specific gravity of 2.62, and was 2.95 per cent of the weight of the water, and would make 2.63 per cent of 70-pound soil, or 2.17 per cent of 85-pound soil. The mud from No. 8 of October 4, when dried, had a specific gravity of 2.61, and was by weight 5.49 per cent of the water, and would reduce to 4.89 per cent of 70-pound soil, or 4.03 per cent of 85-pound soil. 26. The inequalities in the one-day percentages are largely due to the practical difficulty in obtaining uniform portions of the same mixed sample-a difficulty which increases with the higher percentages of silt, which settle faster. Particularly is this noticeable in the larger longer tubes into which the water was poured the fastest, entirely emptying the dipper several times. It is not unlikely, however, that the higher initial reading of the 10-foot tube is to some extent due to the time, though 24 hours, being relatively less, because so high a column requires more time to arrive at the same early stage of settlement. Probably the differences were still more marked at the end of 12 or 6 hours. Certain discrepancies in the readings are due to different judgments of two observers as to average height of mud whose top was sometimes inclined or warped.

27. On account of the difficulty of getting into several tubes water of the same degree of muddiness, a long series of experiments would be required to establish the relationship between the determinations of volumetric percentages of silt, made with different sized tubes; but the above show in a general way that mud in the more slender tubes comes to comparative rest at earlier dates-even in so short a time as three days; and that in the longer tubes the reduction in height of mud from day to day is greater and of longer continuance; the mud, at first a dozen or more times as high in the 10-foot tube, reducing one-half in from 3 to 9 days, while that in the slender boiler gauge tubes may undergo a total reduction of but 10 to 20 per cent. It is also noticeable that with the smaller percentages of mud the tubes of small diameter show but little reduction after the first day.

28. The sampling during 1913 was done about half a mile above the dam site, near the foot of Bowlder Rapids where, in a length of 100 feet and fall of 3 feet, material coming from above, whether carried there in suspension or rolled along the river bed, must all pass, and is thoroughly mixed. This selection of site for sampling was made to eliminate the unknown quantity which results when sampling is done in a quieter stretch of river where material, if any, rolling along the bottom is not caught.

29. The following are the features of the muddiest sample of the year obtained August 16. In a glass tube of 1.15 inches diameter filled to a height of 31 inches the daily readings of silt were in percentages as follows:

Time in days

Per cent

1 2 3 4 5 6 7 8 9 10 31.4 27.4 25.0 24.8 23.2 22.6 22.3 22.2 21.9 21.9

This mud, dried, had a specific gravity of 2.44. Its weight was 6.63 per cent of the weight of the clean water which bore it, and 6.22 per cent

of the muddy water which yielded it. It would produce soil of which a cubic foot in place weighs 70 pounds after being dried, amounting to 6.22 per cent x 62.4÷70-5.56 per cent of the volume of inflow, or 85-pound soil amounting to 4.57 per cent. The voids in the 70pound soil would be 54 per cent, and in the 85-pound soil 44 per cent. The 70-pound soil would be 18 hundredths of the 1-day volume of silt, or 25 hundredths of the 7-day volume. The 85-pound soil would be 15 hundredths of the 1-day volume, or 20 hundredths of the 7-day volume.

30. Smaller volumetric percentages of silt, being of less height and therefore less compressed by their own weight, would require still smaller multipliers to reduce them to soil. But not all silts, though volumetric percentages were the same, would give like multipliers to reduce them to soil, for the character of silt as regards denseness in the tube, specific gravity, etc., is not constant, being dependent on its source and on stage of flood. In general, so far as this year's observations go, the first silt to arrive on a flood is of low specific gravity; and the muddiest water and the higher specific gravities have been found a day or two after the peaks of floods had passed by. 31. To verify the conclusion that silt is not compressed by weight of superincumbent water, several experiments were made of putting equal quantities of the same sample of muddy water into the 10-inch and 10-foot tubes and filling the latter the rest of the way with clean water. There was no material difference found in the heights of the deposited silt.

32. Former silt determinations at San Carlos were made not far from the agency, at or near the gauging station, which was on a straight reach of the river free from rapids, and which is therefore a favorable locality for the heavier silt to have separated from that left in suspension and to be rolling along the bottom, and thus escape being included in the recorded percentages of silt.

33. That the amount of material thus traveling and passing a given point in a specified time is relatively small in this case is known from the fact that such movement is for the most part a slow transit of sand waves on whose upstream gentler slopes the material is rolled up to the summits, whence it descends along the steep natural slopes to the bottom, there to rest till a wave length has passed over; and that such waves in so shallow a river are small; and that their travel is not recorded like that of water and suspended silt, in miles per hour, but in feet per day. The quantity of sand found in the silt at the bowlder rapids was so relatively small as to indicate that not many sand waves per day were reaching that point. But on account of the oft-repeated reference heretofore made to the additional material rolling along the bottom, and the various guesses as to its quantity, experiments were made to better indicate the probable amount.

34. Since silt determinations at the bowlder rapids, where all is mixed, include the material which is presumed to be rolling along the bottom at the old gauging station, parallel observations were made on several days at the two points to show the excess at the bowlder rapids. Because the quantity of silt is fitful, whereby samples taken at short intervals sometimes vary considerably-even as much as 25 per cent the chance of getting abnormal results was avoided by using composite samples consisting each of 8 or 10 samples taken at equal intervals of time, as 1 hour, one-half hour, or one-quarter hour