As to low extremes, it seems well ascertained that the one of 1819'20 was the lowest known prior to 1841—the low depression which succeeded the extreme elevation of 1838, Presumably it was the lowest known during the century. Old Frenchmen of Detroit had no tradition of a level below that of 1819. Statements regarding the stage of the water always make reference to the acknowledged highest and lowest years. Thus we are enabled to fix upon and determine with considerable exactness the relative values of other low periods. The water in 1796 was reported by lake captains to he universally low, and indicating a level five feet below the high extreme of 1838, From that year, they say, it rose rapidly, and continued to rise until 1800, Colonel Whittlesey says: "It was ascertained generally that the water was low in 1790, 1796, 1802, and 1810, Between February, 1819, and June, 1838, there was a continual rise, amounting to 6 feet 8 inches," Old settlers compare the low stage of 1802 with that of 1797, In 1806 it was reported at Cleveland lower than in 1801-02, and declining regularly to 1809-'10. At this date it was reported nearly as low at Buffalo as in 1819. From 1828 it was reported as falling, and in 1833 was 3 feet 10 inches below June, 1838. From this year on we are able to trace the "secular" periods of lake and river with considerable accuracy; and data also exist in regard to other elements which it is proposed to include in our discussion. I give two diagrams, intended to exhibit graphically what is shown more in detail in the tables.
Diagram' No, I shows the curve of high and low water of Lake Erie from 1788 to 1838, constructed in accordance with the above data. In connection with it is given the sun-spot curve, from 1769 to 1838, according to Wolf's tables, reference to which will be made hereafter. The lengths of the periods are also shown, and the lag of the lake periods behind the sun-spot periods.
Diagram No. 2 gives similar data for the term of years from 1834 to 1887, including, in addition to the curves of lake-levels, those of the rainfall and of the temperature (registered at Detroit), and of the sun-spots, according to Wolf's tables.
In these diagrams my endeavor has been to exhibit by curved lines the recurring maximum and minimum periods, eliminating intermediate and irregular fluctuations.
Confining our attention for the present to the curve of rainfall (Diagram No, 2), let us endeavor to ascertain whether among the many and often abrupt fluctuations it is possible to discover any periodicity.
The vertical columns represent years. In the portion devoted to the rainfall variations the horizontal lines represent the number of inches of annual precipitation.
It will be noted that the years 1836 and 1880 were times of excessive rainfall. Between these two extremes, and about equidistant.