CONFIRMATION BY OTHER GAGINGS.

According to the long series of daily gagings at Fort Edward, the computations show a draft of 23 inches, corresponding to 1,775 cubic feet per second on 1,050 square miles, could have been maintained from the Sacandaga reservoir throughout this period with a decidedly smaller depletion than called for by the Mechanicville gagings, and the gagings on the Schroon and the Hoosic tributaries as set forth in the preceding table call for still less storage to maintain 23 inches of run-off, notwithstanding the mean rainfall on Hoosic is about 10 inches less than on the Sacandaga, so that all four of these series of daily gagings made upon the upper Hudson and its tributaries, independently confirm the feasibility of regulating the discharge of the Sacandaga with twenty-five billion cubic feet of storage, so as to provide for a constant rate of flow of upward of 1,775 cubic feet per second. Taking account of differences of rainfall on the respective drainage areas, they conform the estimate of 1,900 cubic feet per second already stated on page 114.

COMPARISON WITH THE CROTON RECORDS OF RAINFALL AND RUN-OFF.

On making a comparison with the Croton records, the direct comparison is somewhat less favorable, until allowance is made for differences in rainfall, forest cover and evaporation. Assuming the run-off per square mile the same, it appears that for the years 1892 and 1895, which were dry years on the Croton, a discharge of 1,500 cubic feet per second could have been at all times furnished from a watershed of 1,050 square miles if storage of twenty-five billion cubic feet was available. In other words, the Croton record would indicate about 200 cubic feet per second less of constant draft would be obtained from an area equal to the Sacandaga drainage in the period from 1898 to 1907, than is given by the Mechanicville record for the same period.

As a warning against the smaller yield that may be expected in years of still smaller rainfall than any since 1888, or in what was probably the dryest year for more than half a century, the long term Croton records presented in the reports of the Croton Aqueduct Commission and discussed in detail in my report on New York's Water Supply made to the finance department in the year 1900, are of especial interest.

These records indicate that during the years from 1880-1882, which years gave the smallest, run-off ever known, a flow of about 1,325 cubic feet per second could have been constantly maintained from a watershed of the same quality as the Croton having an area of 1,050 square miles and twenty-five billion cubic feet of storage available. This is about 200 cubic feet less` than the Croton could have been maintained in the drought of 1895. The Sudbury records of the Boston water works concur wonderfully well with the Croton records in showing the same yield per square mile and the same storage requirement for the dry period of 1880-1882.

There can be no doubt that the yield per square mile of the forest-covered Sacandaga drainage is decidedly larger than for the Croton drainage area. The long term Croton rainfall average is 49.13 inches, for the 40 years 1868 to 1907 inclusive. The Sacandaga rainfall deduced from our preliminary

.

isohyetal map is forty-nine inches. The Sacandaga run-off will be much more in excess than the rainfall, because of the much larger proportion of forest and the higher altitude and the less evaporation.

The weir gagings of the New York Board of Water Supply on Esopus creek for the year 1907 show that the Catskill region gives a yield per square mile 10 per cent. larger than the Croton, although the rainfall at the Croton stations was 15 per cent. larger than for the Esopus stations. The real excess in runoff for this forest-covered mountain region was thus about 25 per cent. by this one year's observations.

The requirements for a water power are less exacting than those for domestic supply, so that independently of the Croton data for periods of drought and after allowing for a smaller evaporation loss, it seems safe to base our Sacandaga water power estimates upon a run-off on a yield decidedly larger than the Croton records of 1,325 cubic feet for the years of greatest drought in the past half century, or than the 1,500 cubic feet per second for the greatest Croton drought of the past twenty years, and to rely upon at least the 1,700 cubic feet per second already mentioned.

YIELD OF SACANDAGA DRAINAGE AREA COMPARED WITH YIELD OF HUDSON. RIVER AT MECHANICVILLE.

Since it is the twenty years' record of daily gagings on the main Hudson river at Mechanicville, where the drainage area is 4,500 square miles, from which we have in the first instance determined the probable yield of the Sacanaaga drainage of 1,050 square miles, it is important to know if the Sacandaga yields more or less per square mile than the remaining 78 per cent. of the Hudson river drainage above Mechanicville.

The most reliable method of determining this is by the comparison of certain series of stream gagings on the respective watersheds. I arranged for these at the earliest practicable date and am under much obligation to the hydrographers of the U. S. Geological Survey for the promptness with which they undertook this work, thereby obtaining a series of comparative measurements of the flow of several tributaries at the close of the remarkably severe drought of August, 1907, and prior to the beginning of the unusual heavy fall rains of this year.

The work of the hydrographic branch of the U. S. Geological Survey in its gagings of streams in New York, had been interrupted by failure of the last Congress to make the necessary appropriation, and as the survey had at its command experienced men admirably qualified for undertaking this work, an arrangement was made with the chief hydrographer by which certain gaging stations upon the upper Hudson and its tributaries were established, this work being paid for out of the funds appropriated by the Fuller bill, to the extent of reimbursing the Geological Survey for its actual expenditure in wages and traveling expenses of the men employed. This arrangement has proved very satisfactory. The gagings have been under the personal supervision of Mr. H. K. Barrows, a hydrographer of much experience. The gage heights have been read daily and frequent current meter measurements have been made for establishing a rating curve at each station. The remarkably heavy rains of September, October and November gave uncommonly good opportunity for obtaining a wide range of rating

measurements. Looking backward, it is now easy to see that check measurements at Indian lake, heights in Schroon and other smaller storage lakes and check gagings at Mechanicville, would have been useful, and these should be taken during the coming year.

We made an earnest effort to obtain an elaborate series of rain-gage records to accompany these streams gagings, but owing largely to a failure to receive the necessary instruments on time, there was an unfortunate delay in beginning the record at the new stations.

The results of the gagings are shown in plate No. 3 and a brief abstract of the comparative results on the several subdivisions of the Hudson watershed is shown in the following table.

It will be noted that the flow of the Schroon is out of proportion to the flow of the other tributaries, by reason of the storage of three feet more or less in depth over the six and seven-tenths square miles area of Schroon lake, and as storage was being accummulated during this period from its watershed of 439 square miles above the Starbuckville dam, some disturbance is caused in its discharge record for these few months, which would disappear in the average of a year or more of daily observations.

The results plainly show that the Sacandaga responds to a rainfall more vigorously than the other tributaries, and that its yield per square mile is decidedly larger at all times, except near the end of an extreme drought.

The flow at Mechanicville at this time after the heavy rains was also thrown very slightly out of direct relation to the natural yield of the watershed, because of the flow from 131 square miles or nearly per cent. of the drainage area above Mechanicville being withheld for the replenishing of storage in Indian lake. Until September fourth, the flow of the Hudson had been increased from time to time by water from Indian lake as is set forth on diagram in plate No. 5.

Daily flow of Schroon is affected by storage in Schroon Lake.

Daily flow of Hudson is affected by storage in Schroon Lake and Indian Lake.

The flow of 775 second feet gaged at North creek on August twenty-ninth came mostly from Indian lake. Probably somewhere near 650 feet of the flow at North Creek, Thurman, Ft. Edward and Mechanicville at this time was coming from Indian lake, and in column six, a deduction, varying from day to day, has been made for this in order to obtain the natural flow.

One of the Indian lake sluice gates happens to have been shut on August 28th to September 1st inclusive. Both sluice gates were wide open at all other times, August 17th to September 4th. This somewhat disturbs the precision of this part of the comparison of gagings on Hudson.

NATURAL YIELD WITHOUT STORAGE IN TIME OF DROUGHT.
For the Sacandaga.

The three independent gagings at Wells, Northville and Hadley given in the preceding table concur in showing that the Sacandaga flow fell to the very low rate of 0.10 or 12 cubic feet per second per square mile in extreme drought, notwithstanding its excellent forest cover; the gagings for the four months following show that after the rains the Sacandaga run-off was increased in much larger degree than the Hudson run-off. A much longer period of comparisons than these three or four months is needed to establish this ratio with precision.

For the Upper Hudson.

The Hudson flow as recorded at Mechanicville near the end of the severe August drought held up unaccountably well in view of the small flow of its tributaries. This fact invites a study in detail during low water of next summer. Its flow including Indian lake averaged about 1,825 cubic feet per second, August 28th to 31st equivalent to 0.40 cubic feet per second per square mile, or deducting 600 cubic feet per second to correct for the storage drawn from Indian lake, gives a natural flow of 0.27 qubic feet per second per square mile. Assistant Engineer Shedd visited Mechanicville to review the record and seek explanation, but no reason is yet found for changing the Mechanicville record for 1907 from that given on page 105 and 120 in the discharge

curves.

The regularity of the record at just the time of our drought gagings is somewhat disturbed by the fact that the paper mills at Mechanicville were shut down from August 20 to August 26th, and that on Sunday, September first, and Labor Day, Monday, September 2nd, the discharge was being held back in storage in the mill ponds.

The Spier's Falls records of flow through turbines is computed from the record of daily electric output, basing this computation on the turbine maker's statement of turbine discharge and efficiency ratio. For August 28th to 31st inclusive, the flow thus computed averaged 890 cubic feet per second, which is probably too small, since it takes no account of leakage, etc. This 890 cubic feet per second including Indian lake storage, is equivalent to 0.32 cubic feet per second per square mile, or deducting 600 feet for Indian lake leaves only 290 cubic feet per second, or about 0.11 cubic feet per second per square mile; also probably an underestimate. These figures are, however, not far from those quoted on page 66 from the census report of 1880, for low water flow, on the excellent authority of Mr. Warren Curtis. So far as can be judged from these incomplete data, the natural yield of the