Figuring on the basis of a 500,000-acre-foot minimum capacity to be maintained and with a dam 180 feet to spillway, desilting would have to begin after 54 years. With a dam 190 feet to spillway, or by use of the 10-foot shutter gates, the life of the project without desilting would be prolonged to 87 years.

On the normal-cycle assumption desilting would have to begin at once to accommodate an annual draft of 300,000 acre-feet, unless spillway gates were provided, in which case it could be deferred 37 years. For an annual draft of 225,000 acre-feet desilting would not be necessary, except during a period as dry as from 1895 to 1904, for 127 years, unless spillway gates were installed, in which case it could be deferred for 164 years. For a period as dry as from 1895 to 1904 desilting would have to begin at once for an annual draft of 225,000 acre-feet, if spillway gates are not installed. With spillway gates raising it to 190 feet, a draft of 300,000 acre-feet per year would be possible a little over half the time and 275,000 all of the time, if silt removal begins as soon as its accumulation becomes enough to make its removal economical.

On account of the necessity of storing two or three years' supply, a reservoir impounding 714,450 acre-feet with a dam 180 feet high to the crest of the spillway was chosen. The largest flood of which there are definite measurments (December, 1914) was used in determining the size of the spillway. To make sure that the spillway was large enough, the discharge was doubled. This flood could have passed a 300-foot spillway with a depth of 11 feet. Placing 10-foot gates in the spillway would reduce the clear space to 240 feet and increase the depth of the water over the spillway to 14 feet. Doubling the duration of the flood as well as its intensity increased the depth to 17 feet, so a dam 200 feet high over all should be sufficient. The use of the spillway gates will increase the capacity of the reservoir to 854,810 acre-feet.

Regardless as to whether we use a 200,000 acre-foot or a 300,000 acre-foot draft, the diagram shows plainly the necessity of providing ample hold over capacity, and it also appears that 500,000 acre-feet, as assumed by the Army Board, should be the minimum capacity of the reservoir.

To provide a reservoir of this capacity and at the same time to defer desilting operations for a reasonable length of time, a dam capable of storing water to the 190-foot contour would be the most suitable height. This capacity could be obtained by means of a dam having the same height, 180 feet to spillway, as recommended by the Army board, but with the addition of 10-foot shutter gates along the spillway.

ANNUAL DRAFT.

In the case of the San Carlos project, the quantity of water that may be considered the proper annual draft from the reservoir is a problem affecting the operation of the entire project, and may well be a subject upon which opinions might widely differ.

The Army board has based its calculations of the annual draft on the critically dry period; which means that there is a full water supply guaranteed to the lands during a period as dry as the driest on record. On account of the protracted dry period on which we

have the records of the Gila run-off, the quantity of water to be delivered each year by the reservoir must be made correspondingly small and the assured draft from the reservoir must be made considerably less than the normal or average supply. While it is proposed to utilize as much of the overflow from the reservoir as is possible, yet the area of the land to come under the project must necessarly be small and a project based upon a minimum supply can not, under more favorable conditions, be expected to make the best uses of the normal supply.

By basing a project upon a draft representing more nearly the normal supply, we approach a more thorough utilization of the available supply, less losses by evaporation and overflow, and a larger area may be supplied under the project, but it is impossible to guarantee a full supply of water to this area during a period of extreme drought similar to the driest on record.

It should be remembered that the dry period of record to which we have already frequently referred, viz., from 1899 to 1904, was a remarkable one, not only on account of small amount of the run-off of the Gila River for the different years, but more so on account of the continued succession of dry years.

The full possibilities of the project would not be developed were it decided to limit the draft to the quantity of water necessary to insure against future deficiency in case of a repetition of the worst period of which we now have record. It is believed that the draft should be based upon an amount more nearly equal to the normal basis.

The Salt River project is operated on the normal-supply basis. It should be pointed out, however, that the Salt River project differs from the San Carlos in this respect, that under the San Carlos project all of the lands are assumed to have the same right to the water, whereas under the Salt River project certain lands, as provided for by the Kent decree, have a preferential right, and these preferred lands during years of shortage receive a full supply, while the remaining lands get only what is left, or must be supplied by means of pumps.

The Salt River project must also conform with the provisions of the reclamation act.

The Salt River project as recently recommended1 is to embrace 195,000 acres. Seventy thousand acres of class A, or preferred, land are to be supplied with gravity or stored water and the remaining 125,000 acres to be supplied by gravity and stored water, supplemented by 30 pumps.

The mean anual run-off of the Salt River at Roosevelt for the period from 1889 to 1911 was 808,000 acre-feet, with a maximum flow of 3,260,000 acre-feet and a minimum flow of 154,000 acre-feet." The duty of water as recommended by the present board of survey above referred to, was assumed at 41 acre-feet at the point of diversion and 3.3 acre-feet on the land.

1 Report of a board of survey convened by the Secretary of the Interior to consider the selection of lands to be included in the Salt River project, Arizona, consisting of F. W. Hanna, supervising engineer, Reclamation Service, and W. A. Farish and Frank H. Parker, representing the Water Users' Association. Report published Jan. 15, 1914. * Eleventh Annual Report of Reclamation Service, p. 45.

While the difference between the mean and the minimum flow of the Salt River has been relatively greater than that recorded for the Gila, yet the protracted dry spell recorded for the Gila has been more severe than any similar period on the Salt. Basing these calculations on the critically dry period of record on the Salt River, but assuming that supplemental pumps were installed, the board of survey in their report show that the water supply for the lands under the Salt River project would have been less than the mean during 6 of the 25 years of record and that the greatest deficiency would have been about 60 per cent below the normal supply. With respect to the deficiency during the driest periods, it is believed that the San Carlos project, as outlined above, compares favorably with the Salt River project.

Considering all phases of the annual draft problem, it is believed that it would be better to reap the benefit of the normal water supply during a greater portion of the time and to assume the risk of failure in unusually dry years than to provide for a much more limited use of the water at all times in order to insure against possible droughts during exceptional years. As pointed out before, the period extending from January, 1902, to August, 1904, during which the reservoir would be empty, was one of exceptionally low precipitation over the West, and from records available at other points may not be reasonably expected to occur more frequently than once in 30 or 40 years.

The permissible annual draft of a practicable reservoir at San Carlos, based on our present knewledge, should therefore lie between 300,000 acre-feet as an upper and 250,000 acre-feet as a lower limit. A greater draft would not be permissible, while a smaller draft would fail to utilize the possibilities of the project. In so far as the annual draft is concerned, these two figures form the basis of the maximum and minimum cost of the project.

In the computation of the annual draft, the beneficial effect of the flow of the San Pedro has not thus far been considered. Sufficient measurements have not been taken to determine the flow of the San Pedro with any degree of certainty. The United States Army board assumed its discharge as one-tenth that of the Gila. The observations made during this investigation seem to indicate that the flow varies from one-fifth of the flow of the Gila River during low water to one-tenth during flood periods.

For lack of better information, the flow of the San Pedro has been assumed to be one-tenth of the flow of the Gila. The San Pedro discharges into the Gila about 42 miles above Florence, and its flow during flood periods would decrease the amount necessary to be drawn from the reservoir for irrigation, permitting a larger amount to be stored. The effect of this flow on the behavior of the reservoir is shown in the mass diagram, plate No. 72, in volume of maps, accompanying this report, and is based on an annual 5,000 acre-feet transmission loss in the river between the dam site and the point of diversion above Florence. The effect of adding the flow of the San Pedro is to shorten the empty reservoir period, so that instead of going dry February 1, 1902, the water supply would have held out under a 300,000 acre-foot draft until December 1, 1902. Consequently, by adding the flow of the San Pedro, the reservoir with a capacity of 714,450 and the above draft would have been dry from December 1,

1902, to August 1, 1904. With a capacity of 854,800 acre-feet, under the same conditions, it would have been dry from May 1, 1903, to August 1, 1904.

During the succeeding cycle, the effect of the San Pedro would have been nil, owing to the discharge of surplus water over the spillway. The above calculations have been based on the amount of water actually available. Similar calculations, using the amount physically available, show that the only effect of using the water now diverted for irrigation in the Solomonville Valley is to diminish the time the reservoir is empty during the extremely dry cycle. The mass diagram, just referred to, also shows the behavior of the reservoir, including the effect of the San Pedro River, and water now diverted in the Solomonville Valley.

The mass diagram showing the effect of diversions at Solomonville is based on an annual net loss of 58,000 acre-feet. This figure is explained later in Appendix B of this report. This diagram indicates that the 714,450 capacity reservoir with an annual draft of 300,000 acre-feet and with combined flow from Solomonville and the San Pedro would have been empty from March 1, 1904, to August 1, 1904. With a capacity of 854,800 acre-feet it would have contained 2,000 acre-feet on August 1, 1904, when the succeeding flood began. During the balance of the 21-year period the addition of the Solomonville diversion water or the San Pedro would have no appreciable effect, since the run-off without this supply was more than sufficient to meet the demands of the project.

HISTORY OF IRRIGATION ON GILA RIVER.

Irrigation, an essential requirement of agricultural development in the Southwest, has been practiced along the Gila since prehistoric times. Remains of ditches and other unmistakable evidences of irrigation by an ancient race, which apparently reached a higher stage of civilization than the Indians known to the early settlers, are to be found at different places in the valleys of the Gila and its tributaries.

The Pima Indians, who may be descendants of these prehistoric irrigators, were irrigating a large area of land along the Gila when the first exploring Jesuits penetrated this territory during the middle of the sixteenth century.

While the Jesuits and their followers intermittently irrigated a limited area of land along the Santa Cruz, and possibly along the San Pedro (both tributaries of the Gila), yet no important irrigation other than that practiced by the Indians took place until some time after the American military occupation in the middle of the nineteenth century, when protection to the settlers and markets for the crops were afforded.

A small area of land was irrigated at this time, in connection with the overland stage stations. This stage line was established in 1857. and the route, after leaving Tucson, was through the Pima villages on the present Gila River Indian Reservation, cutting across the big bend of the Gila and touching the river again at the present Gila Bend station; thence continuing along the river to its confluence with the Colorado at Yuma. All of the land formerly cultivated in connection with these old stage towns, as well as the other older irri

gated area on the lower Gila, has since been allowed to revert to the desert state.

Subsequent diversion of the waters of the Gila by the whites proceeded in a rather uniform manner up the stream. Irrigation in the Florence district started earlier than in the Solomonville Valley, while the first irrigation in the Duncan Valley took place at a later date than in the other districts lower down on the Gila.

The first irrigation in the Florence district was initiated by, and for the profit of, the early Indian agents of the reservation, just below and contiguous to the Florence district.

The pioneers of the next valley up the river, the Solomonville district, were Mexicans, who ventured into this uncivilized territory in the early seventies. Americans followed, and with the advent of the Mormon colonists a few years later rapid agricultural development took place.

In the Duncan district, as well as along the upper tributaries, the Mexicans made the earliest diversions. Irrigation here, however, had many disadvantages, adverse natural conditions together with remoteness from civilized centers or military protection considerably delaying later diversions of the waters of the Gila for irrigation.

The progress of irrigation along the Gila River since the American occupation has varied in the different districts. Coincident with the development and consequent diversions of the river waters in the valleys on the upper reaches of the stream, the water supply for the irrigated districts lower down soon began to suffer, and in the extreme lower valleys many acres of irrigated land were no longer cultivated. In the middle Gila districts the agricultural development was either stopped entirely or greatly retarded.

The present status of irrigation along the Gila is summarized in the following tabulations.

The areas shown were actually under irrigation by the use of ditches and pumps, taking water direct from the Gila River at the time (1914) of the survey made in connection with this investigation.

24, 045. 0

No. 5. Lower Gila district (west of Pima Indian Reservation).