best exhibit of what would be expected to be the muddiest seasons at San Carlos; hence, the daily volumetric percentages from the one record, and the corresponding daily discharges from the other record, are here reproduced in the following table:

TABLE XII.-Sediment observations at San Carlos, in 1904.

"Means for month" are derived by dividing the sums of the daily discharges into the sums of the products of such discharges multiplied by their corresponding percentages of sediment.

66. The reduction to 70-pound soil is made by dividing by 5, as indicated in Table V. Considering that Table V is based on water columns 31.25 inches high, while in Table XII, 100 cubic centimeter test tubes, which are much shorter, were used, the divisor 5 is doubtless too small for application here, but the difference is on the side of conservatism. It is probable that the July percentage would have been smaller had the river been running throughout the month. 67. The sediment data for 1905 is found on page 41 of Water Supply Paper No. 274, which gives determinations in milligrams per liter, or parts in a million by weight, of suspended matter in 30 composite samples taken between April 9, 1905, and January 2, 1906. The volumetric percentages in samples taken morning and evening from March 22 to December 22, 1905, read after 24 hours settlement in 100 cubic centimeter graduated glass tubes, are in unpublished record, of which

a copy has been kindly furnished by the Phoenix office of the Reclamation Service. The river's discharge at San Carlos during a portion of the year, covering the muddy months and more, is in an unpublished record furnished by the United States Geological Survey. From these three sources, the following tables are compiled.

68. First, a comparison will be made between the 1-day volumetric percentages and the percentages by weight, for the gentleman who did the sampling has written that of the same sample well mixed, one part was subjected to the volumetric test and the other was sent to the laboratory for analysis.

TABLE XIII.-Comparison of volumetric and weight percentages of silt in 1905.

Apr. 9, 10, 11, 12..
Apr. 17, 18, 19, 20.
Apr. 24, 25, 26, 28, 29.

Apr. 30; May 3, 4, 5, 6.
May 10, 11, 12.

May 14, 16, 17, 18, 19.

May 22, 23, 24, 25, 26, 27.

May 28, 29, 30, 31; June 2.

June 9, 10, 11, 12, 13, 14, 15.

June 18, 19, 20, 21, 22, 23, 24.

June 25, 26, 27, 28, 29, 30 July 1.
July 2, 3, 5, 6, 7

July 9, 10, 11, 12, 13, 14, 15..
July 16, 17, 22.

July 23, 24, 25, 26, 27, 28.
July 29, 30, 31; Aug. 12..
Aug. 13, 14, 15, 16, 17, 18, 19..
Aug. 20, 22, 23; Sept. 5, 7..
Sept. 10, 16, 21; Oct. 1, 2, 6, 7.
Oct. 8, 9, 10, 11, 12, 13, 14..
Oct. 15, 16, 17, 21, 22, 23, 24..

Oct. 25, 26, 30, 31; Nov. 2, 3, 4.
Nov. 5, 6, 7, 8, 9, 10, 11.
Nov. 12, 13, 14, 15, 16, 17, 18.
Nov. 19, 20, 21, 22, 23, 24, 25.
Dec. 5, 6, 7, 8.

Dec. 10, 11, 13, 14, 15, 16.
Dec. 17, 18, 19, 20, 21, 22, 23.
Dec. 24, 25, 26, 27, 28, 29, 30.
Dec. 31; Jan. 1, 2 (1906).

Mean...

69. The above column of ratios is, apparently, so erratic that further inquiry is needed before passing judgment, particularly since similar ratios in Table V lie within much narrower limits; hence reference is made to El Paso determinations to be found on page 400 of the Third Annual Report of the Reclamation Service, where percentages by volume after standing 7 days, and percentages dry, are given in parallel columns, which, by dividing, give ratios as follows: 0.46; 0.25; 0.24; 0.24; 0.28; 0.28; 0.26; 0.26; 0.22; 0.25; 0.24; 0.32; 0.29; 0.32; 0.34; 0.33; 0.34; 0.28; 0.19; 0.27; 0.25; 0.30; 0.15; 0.30; 0.19; 0.24; 0.62; 0.33; 0.92; 0.41; 0.26; 0.46; 0.29; 0.33; 0.19; 0.34; 0.29; 0.14; 0.30; 0.13; 0.15; 0.50. The smaller ratios generally attach to the smaller percentages of silt, but the largest ratios are undoubtedly erroneous. Rejecting all that are above 0.33, the average of the rest is 0.25, which would reduce to 0.20 had the readings been, as in Table XIII, one day instead of seven days. The 0.20 must be still further reduced to apply to Table XIII, for the average seven-day volu

metric reading at El Paso, from which it is derived, was about 7 per cent, or an equivalent one-day reading of nearly 9 per cent, which, being more than twice the average in Table XIII, would represent silt more compressed under its own weight. Hence the larger ratios in Table XIII, say, 0.10 to 0.17, are not unreasonable, when applied to the relatively smaller amounts of silt in 1905.

70. It is considered that the percentages by volume in 1905, observed by the sampler who knew the condition of the river each day, are more reliable than the published percentages by weight recorded in Water Supply Paper No. 274, for the latter may have been more in the nature of a by-product in analyses intended more for the determination of chemicals than of amount of silt. Hence the unpublished volumetric data will be used for 1905, for somewhat higher quantities of silt will thus appear, which in the present study are the more conservative to adopt; and the highest ratio in Table XIII, namely 0.17, will be used to reduce the volumetric percentages of that year to weight of dry matter, or 0.17 <62.4÷70 0.15 will be used to reduce to volume of 70-pound soil. 71. The volumetric percentages and discharges are compiled in the ollowing table:

=

TABLE XIV.-Sediment observations at San Carlos in 1905.

Means for month in the above are derived by dividing the sums of the daily discharges into the sums of the products of such discharges multiplied by their corresponding percentages of sediment.

1 Average gauge height: Apr. 1 to 8, 13.5; Apr. 9 to 30, 14.

*May 14 to 27 averaged about 1 per cent. See Table XIII.

'Average gauge height: May 1 to 13, 12.7; May 14 to 31, 11.6.

TABLE XIV.-Sediment observations at San Carlos in 1905-Continued.

1 Dec. 23 to 31 averaged about 0.5 per cent. (See Table XIII.) For reasons given above in paragraph 70, soil in place, weighing 70 pounds dry, is taken to be fifteen one-hundredths of the one-day volumetric percentages in the year 1905.

72. If each percentage in Table XIII be applied to the corresponding average discharge of the days represented by the composite sample, as far as such discharges are given in Table XIV, the monthly percentages of 70-pound soil become as follows: April, 0.74; May, 0.22; June, 0.17; July, 1.47; August, 1.29; September, 1.74; October, 0.20; November, 0.47; December, 0.09. But, as stated before, the percentages derived in Table XIV will be used in preference to the above, for they are based on fuller data and are higher.

73. So far as known, no further observations were made at or near San Carlos till those of 1913, appearing in Table IV. These have not the river discharges corresponding to them, but they can be made of use. Their averages are: For August, 7.84 per cent; September, 5.67 per cent; and October, 2.81 per cent. Since these months are in the season of muddy river with fluctuating flow, the percentages of silt with respect to the flow is known to exceed the above means of the daily readings; for in this period the little rises are accompanied by higher percentages of silt. Were the percentages of silt proportional to the discharge, then the quotient of their sum divided into the sum of their squares would be the correct result. But the general observation during 1913 has been that, though floods during the

muddy season bring mud, the percentage falls short of increasing as rapidly as the flow; for often the smaller rises carry higher percentages than the larger floods, and generally the muddiest water has been found after the peak of the flood had passed. Taking all the muddy months that have but few, if any, clear water days, and whose records are complete, it is found that on the average their percentages of mean flow representing sediment approximate more closely to the square root of the mean of the squares of the daily percentages than to the quotient of the sum of the squares divided by the sum of the daily observations, as is shown in detail in the following table:

Were the daily percentages of silt observed at San Carlos during August and September used in connection with the daily discharges of the river at Kelvin one day later, of which the record has been since received, the resulting mean percentages would be 11.5 for August and 7.3 for September.

74. The flood of October, 1913, fell far more slowly than the percentage of silt decreased. In view of this fact, and of the averages in the above table, it is judged that the most probable percentages for August, September, and October of 1913 are, respectively, 11.20, 8.60, and 8, which, considering that long test tubes were used, reduce to 2.24 per cent of 70-pound soil for August, 1.72 per cent for September, and 1.60 per cent for October. The silt from January to June, 1913, varied, as before stated, from a stain to three-eighths of an inch in the water barrels, or from zero to 1 per cent, making the equivalent of 70-pound soil to average not more than 0.10 per cent during that time. There was but one flood in June; and July had fewer rises than August.

75. Having now exhausted all known data, the results are summarized in the following table, in which the percentages represent soil of which a cubic foot in place would, if dried, weigh 70 pounds, excepting those of 1895, 1899, and 1904, deduced by dividing volumetric percentages by 5, which are now thought to represent a somewhat lighter soil, but are nevertheless used without reduction, since in this analysis all uncertainties are resolved in favor of the higher quantities.