Dividing this total as before, we get as an approximate average of the heavy rains in the watershed 174.24 inches. Total number of days with stages of 20 feet or over in the river..... 141 Dividing as before, we get as the probable amount of rainfall in the second period required to produce a day of 20 feet or over in the river, 1.236 inches, as against 1.220 inches in the first period. The difference between the line of reasoning employed in getting the above results and those given in Water-Supply Paper No. 234 is that the author of the latter attempts to differentiate between the stages produced by rains of varying intensity and to assign to such rains a given number of so-called "flood days," while in this paper the assertion is made that in the first period there were a given number of days with a stage of 20 feet and over in the river, and that during that time the heavy or flood-producing rains amounted to so much. Dividing, then, the total of the flood-producing rains by the corresponding number of days with a stage of 20 feet or over, the results given above are reached, viz, that for the first period it took 1.22 inches of rainfall to produce a day with a 20-foot stage in the river. These figures contradict the contention that an equal depth of rain in the last period as compared with the first produced more severe floods in the river. I only present them to show how easy it is to arrange data so as to prove both sides to a question. While this line of inquiry is open to less objection than that followed by Leighton, it does conform to the plan of the latter in so far as it uses the number of days that the river stood at or above 20 feet, instead of taking into consideration the actual height of the water. The most that can be said is that this form of inquiry shows no increase in flood intensity. Rainfall and run-off of the Ohio Basin.-We now come to a different and more reliable form of investigating this question of the relation of precipitation to run-off. We have no direct method of measuring the run-off, but we can reach a fair approximation to it by a comparison of the rainfall and river data for any given watershed. If, for example, the surface conditions over any considerable part of a watershed have been materially changed by deforestation or other means, and if, as claimed, such change operates to increase the run-off, then the flow of water in the streams after the change has been brought about should be greater for equal depth of precipitation. This method is a rough one, to be sure, but it appears to be the only one permitted by the records as they exist. Cincinnati, Ohio, has been chosen as the point whose river observations are best adapted to our purpose, although some objection to that place lies in the constriction of the natural river channel caused by the encroachment on the banks of the stream by various artificial structures. The station at Pittsburg, Pa., is better situated for comparative purposes, but the low-water stages at that place of late years have been vitiated by the construction of the Davis Island dam. The construction of dams at several places in other rivers has lowered the value of low river gauge readings for comparative purposes. In the tables which follow I have given the actual mean monthly stage of the Ohio at Cincinnati for every month of the period 1871 to 1908. The average of these monthly means has been computed for the first period of nineteen years; these averages have been summed up for the twelve months of the year, and that sum has been divided by twelve in order to get the annual mean. The number so obtained, 17.3 feet, is therefore the average stage of the river for the entire nineteen years, as computed from all of the daily stages for that period. In like manner the average stage of the river for the second period of years has been computed and is given in the following table: Mean monthly and annual river stages in the Ohio River at Cincinnati, Ohio, for the period 1871-1908. Year. [In feet and tenths.] Jan. Feb. Mar. Apr. May. June. July. Aug. Sept. Oct. Nov. Dec. Annual. 1878. 20.0 24.2 24.2 14.5 22.0 12.5 1879.. 1880.. 29.2 28.3 35.2 23.2 26.6 32.8 22.9 11.4 7.1 24.6 1881. 20.7 33. 1 9.0 10.6 1896. 14. 1 23.6 22.5 26. 1 12.3 6.5 7.5 6. 1 4.9 3.0 3.4 9.4 1897. 14.4 36.3 40.3 26.3 17.4 13.6 8.4 8.0 5.1 3.9 6.4 13.4 4.3 10.3 19.2 15.8 The average precipitation for the watershed has not been so easily obtained. Only in exceptional cases are continuous measurements of precipitation available for comparative studies. The government records in large cities are of necessity made from gauges whose imme diate environment has been changed repeatedly in the course of a long series of years, and it was for this reason that the rain-gauge records from Cincinnati and Pittsburg were ignored. The points selectedviz, North Lewisburg and Portsmouth, Ohio, and Confluence and Franklin, Pa. are the best and practically the only long-period records available in this watershed. A better distribution throughout the watershed would have been preferred, but it is not possible to obtain it. The precipitation, like the river stages, has been computed in periods of nineteen years each. The tables follow: Annual precipitation in the Ohio watershed for the period 1871 to 1908, inclusive. Average precipitation in the Ohio watershed, as determined from the stations above named: Inches. 1871 to 1889. 41.3 1890 to 1908.. 41.8 I consider that the results secured from the discussion of the precipitation and the gauge readings in the Ohio basin, as given in the foregoing tables, form one of the most important contributions made by this paper. Here we have avoided the using of indefinite and meaningless data, and have taken the longest period of time for which accurate records can be secured on a watershed that is suitable for this line of inquiry. We have not simply counted the number of days that the river stood above some arbitrarily selected stage without taking into consideration the exact height of the river. Neither have we divided the gauge readings by some arbitrarily selected portion of precipitation data. On the contrary, we have endeavored to profit by the errors of previous investigations, and to lay the foundation of an inquiry that would mean something when we reached the end of our computations. For this reason we have selected a typical station on the main stream that drains the Ohio Basin and have discussed rainfall data that are the most accurate of any in the region, having been subject to less errors due to varying environments. Any deductions made from an inquiry founded with less care, or from data of a less degree of accuracy, must bring results from which it would be unsafe to form definite conclusions. Now let us see what is the result. The average stage of the river for the first nineteen years is 17.3 feet, and for the last nineteen years 17.5 feet, showing that there is practically no change in the run-off of the Ohio Basin between the first period and the last. When we examine the average precipitation over the watershed that is drained by this river we find that for the first nineteen years it was 41.3 inches, and for the last nineteen-year period it was 41.8 inches, a slight increase in precipitation for the latter period that agrees precisely with the slightly greater average flow of water. There is a perfect agreement here between the precipitation and the flow of the stream. I do not know what has been the area deforested in this valley during the thirty-eight years under discussion, but whatever it is it seems to be apparent that such altering of the relation of forest area to cultivated area has had no appreciable effect on the flow of the Ohio River. I am aware of the fact that by the studying of short periods of data on small tributary streams, and especially by the grouping of data dissimilar from what is employed in this discussion, all manner of results may be shown. I believe that the reader will acknowledge that I have shown in the several preceding paragraphs that the average discharge of the Ohio River, where I presume deforestation has been as great as in any other part of the country during recent time, has not changed for a period of thirty-eight years, except as caused by precipitation. It will now be interesting to know how the two periods compare with regard to extremely high water and extremely low water, and this will be discussed in the coming pages. High water and low water on the rivers of the Ohio basin.—I had Prof. H. C. Frankenfield, Chief of the River and Flood Division of the Weather Bureau, compile the data from one station on the Cumberland, three on the Tennessee, and five on the Ohio, and establish the average high water for the four wet months, January to April, and the average low water for the four dry months, July to October. He then 26320-10-3 took the departure from the normal, both for the precipitation and for the height of the rivers, and found that the average high water was no higher and the average low water was no lower for the last half of the period than for the first half. The differences were so slight as to be inappreciable, but what changes occurred were in favor of the low water being slightly higher and the flood waters slightly less. There were variations in the periods and intensities of floods that bear a direct and proper relation to the precipitation. In making his report, Professor Frankenfield points to the fact that the low-water stages at Pittsburg, Pa., and Nashville, Tenn., are not fairly comparable with those of the other stations on account of permanent pool stages caused by dams operated during the low-water season for purposes of navigation. The first dam below Pittsburg was placed in operation in 1885, and that at Nashville in 1904. The effect of these dams is to furnish higher low-water stages than would result without them. The effect upon the normal low-water stage at Nashville was not marked, but at Pittsburg it was perceptible. However, in his conclusions he did not make allowance for the slightly higher low-water stages at Pittsburg on account of the dam, but when included with the other stages of the river this defect probably is not apparent. According to our line of reasoning, which we believe to be fair and conservative, it is shown that the average discharge of the Ohio River is not greater as the result of deforestation during the last nineteen years than during the preceding like period, and that the average high water in the rivers of the entire basin, which includes the Tennessee, the Cumberland, and the Ohio, is not higher and the low water is not lower. Are real flood stages more numerous than formerly? The next line of inquiry will be for the purpose of determining whether or not there has been in recent time an increase in the number of days that these rivers were at or above the flood stage, and in making this inquiry exact flood stages will be used, not simply gauge readings less than flood. Again I called on Professor Frankenfield to prepare the necessary data. As the data was not complete with regard to flood stages for the first ten years of the period that we have been discussing, he took a period of ten years less in length, beginning with 1879, and as the result of his computations we have the following table |