bore less than the average amount of silt, the mean soil to water ratio for the two years is given as 0.445 instead of 0.626. Because drier years have generally higher percentages of silt the 0.445 is on page 162, raised to 0.536 per cent to represent 85-pound soil over a term of 12 years.

49. The most exhaustive presentative of the Rio Grande silt determinations is the Department of State publication entitled "Silt in the Rio Grande," by W. W. Follett, giving observations at El Paso from May, 1897, to June, 1910, and at San Marcial from August, 1905, to January, 1913, all in the most approved form of percentage by weight dry, together with the corresponding discharge of the river. The highest percentages by weight given for El Paso are 10.50, on page 27, and 10.44, on page 28, which are considerably in excess of the highest recorded on the Gila River. The San Marcial table, however, on pages 31 to 52, contains many higher percentages-even up to 21.36 per cent by weight. It may well be doubted that so high a percentage can be carried throughout the cross section of the river on the grade it has at San Marcial, but such sampling may be sometimes possible next the bottom of a river, which will show too much, not only because it is too silty to represent the average of the river at the time, but also because it does not travel as fast as the river. However, the large number of high percentages, when compared with the findings for several years on the Gila, leave hardly a doubt that the Rio Grande at Elephant Butte is muddier than the Gila at San Carlos. 50. The general average for 16 years at San Marcial, deduced by Mr. Follett, is that the volume of soil in place, weighing when dried 53 pounds per cubic foot, will be 1.66 per cent of the water supply. Substituting the 70 pounds and 85 pounds herein used for Mr. Follett's 53 pounds the 1.66 per cent becomes 1.26 per cent to represent surface earth to be periodically dredged, or 1.04 per cent to represent old deep deposits to fill the reservoir.

51. Another important fact easily deduced from Mr. Follett's very complete tables and useful alike at San Carlos, since the physical conditions are similar, is that the variations in percentage of mud are pronouncedly seasonal, for by averaging the monthly percentages given by Mr. Follett on pages 61 to 64 there results the grand averages tabulated below:

TABLE VI.-Average monthly percentages of soil carried in suspension by the Rio Grande at San Marcial, N. Mex., in 16 years.

In the last column of the above table, the four muddy months average 2.63 per cent and the eight clean months average 67 per cent. Special attention is called to these results, for they help to support the similar though smaller figures to be compiled from the Gila River data, particularly the segregation of the year into two seasons-the very muddy and the comparatively clean.

52. The following is a compilation of all known Gila River silt data: 53. In Water-Supply Paper No. 33, on page 36, is a quotation from Mr. Arthur P. Davis's report of 1896, as follows:

Attempts have been made, however, by private parties from time to time to measure the amount of suspended material. The most important and reliable of these measurements were those made by Mr. Albert T. Colton, in the month ending August 7, 1893, and by Mr. W. Richins, at the Buttes. Mr. Colton found the percentage of silt by volume averaged 2.2 per cent. As these observations were taken and reduced by Mr. Colton, who is a competent engineer, they are adopted as correct. The observations by Mr. Richins extended from July 29 to December 31, 1895. They were taken by the following method:

A sample of the water was poured into a slender glass tube until it reached the height of 100 divisions on a convenient scale, and was then allowed to settle several days, until the main portion of the water was clear. The height of the sediment on the same scale was then read and the result recorded as the percentage of mud carried by the water. The existence of this record and the ease with which more of such observations could be taken, made it important that an approximate relation be established between the volume of this mud and that of the actual solid matter it contained. For this purpose several laboratory determinations have been made with muddy samples of Gila water by settling and reading the volume of mud as above and then drying the residue at 100° C. and determining its volume.

These observations indicate an average ratio of dry matter to mud of about one-fifth, and this factor has been used to reduce the mud observations of Mr. Richins to solid matter.

It should be remembered that these observations take no account of material that is rolled on the bottom of the stream.

54. By "solid matter" and "dry matter" in the above is meant earth which would fill a reservoir, as is plainly shown by the use of it in the report. In another place the same is called "solids."

55. In Paper 33, on pages 37 and 38, are the observations by Mr. Richins, in tabular form, from which monthly totals are taken as follows:

TABLE VII.-Silt observations at The Buttes in 1895.

56. Further observations, made and reduced in the same manner as above, were taken at The Buttes from January 1, 1899, to July 31, 1899, and tabulated on page 39 of Paper 33, giving monthly totals as follows:

TABLE VIII.-Silt observations at The Buttes in 1899.

57. The above Tables VII and VIII taken together, though for different years, cover a year and three days and show four muddy months and eight clear ones.

58. Valuable silt observations were made at the head of Florence Canal below The Buttes from November 28, 1899, to November 5, 1900, and recorded in Bulletin No. 44, University of Arizona Agricultural Experiment Station, by R. H. Forbes. The record, as given on page 185 of the bulletin, is by weight in parts per hundred thousand. The recorded determinations are of composites each of about a week's sampling, and the corresponding river flow is not given. Hence only averages of analyses made each month will be used, for the nearest river gauging during the time was at San Carlos, and it is known that floods in the San Pedro materially affected the results. The monthly averages are as tabulated below:

TABLE IX. Silt observations below The Buttes in 1899-1900.

The high percentages in August and September are attributed largely to San Pedro floods of that year by Prof. Forbes in accounting for relatively large percentages of black alkali found those months, which has its source largely in the granite rocks of the San Pedro watershed, and as stated on page 191 of the bulletin, the weather maps showed the San Pedro to be largely the origin of those particular floods. Hence the percentages for August and September would have been considerably smaller had they been determined above the

San Pedro at San Carlos. On page 188 of the bulletin Prof. Forbes refers to the San Pedro as "perhaps the muddiest tributary of the Gila."

59. A more definite opinion on the relative amounts of silt at The Buttes and at San Carlos appears in the First Annual Report of the Reclamation Service, by F. H. Newell, chief engineer, which, on page 83, states as follows:

Observations at The Buttes reservoir site on Gila River have revealed such high percentages of solid matter as to render the maintenance of a reservoir at that site expensive. Investigation has shown, however, that a large proportion of the silt received by the Gila waters comes from the basin of the San Pedro River

*

* *

The San Carlos Reservoir is located above the San Pedro Basin, and while it intercepts the water supply furnished by the headwaters of the Gila proper—the San Francisco, Eagle Creek, and other streams flowing from the lofty Mogollon Mountains, which furnish very little solid matter to the reservoir-it escapes the silt-laden waters of the San Pedro which joins the river below.

From these considerations it is believed that less sediment will be encountered in the waters of the Gila at San Carlos than at The Buttes. Observations have been made which confirm this expectation.

60. The earliest known record of silt determinations made at San Carlos is that by the United States Geological Survey in 1901, kindly furnished by the Phoenix office of the United States Reclamation Service, which is in tabular form as follows:

The percentages in the record are only to tenths. The 0.04 per cent for Decem er is taken from page 82 of First Annual Report of the Reclamation Service. January, February, and March, not given above, are known to be very low in percentage of silt.

61. Sediment observations at San Carlos in 1902 are published in First Annual Report of the Reclamation Service on page 82, in form as follows, excepting the last two columns:

TABLE XI.-Sediment observations at San Carlos in 1902.

1 Water clear February to June.

The 29 in third column is 39 in the record, owing to error in addition as seen on detailed observations received from the Phoenix office of the Reclamation Service.

* Water clear in October and November.

62. It is to be observed that in this table the run-off is multiplied by the percentage of dry material to give acre-feet in column seven, from which it is inferred that a cubic foot of soil in place was assumed to weigh, after being dried, the same as water, or 62.4 pounds. The last two columns are now added to include soils at 70 pounds and 85 pounds, for use in this paper. On examination of discharge records it appears that some of the months called "clear" in the above silt record, are reported dry, namely, April, May, June, October, and November.

63. No record has been found of silt observations, if any, made in 1903.

64. For the year 1904, an unpublished record has been furnished. by the Office of Director of United States Geological Survey, "showing comparative turbidity and sediment determinations made of the water of Gila River at San Carlos, Ariz., by R. H. Ross under instructions of Sheldon K. Baker, hydrographic aid, from July 21 to October 24, 1904." The record shows that determinations of sediment were made "by allowing sedimentation for 24 hours in a 100 cubic centimeter graduate."

65. These determinations made by Dr. Ross are especially valuable in showing results occurring during the first months of ordinary flood flows following 5 years of drought, and also following a period of dry river, for the discharge record shows the river dry from May 20 to July 20, and nearly dry for months before. Such dryness of the land and the river bed followed by the numerous moderate floods, as shown by the discharge record, are expected to give the highest percentages of silt. The completeness of these records as to percentage and flow, both made by the same trained mind, makes the