1. A possible lunar tide; but so small and so broken in upon by greater causes as to be of very uncertain value.
2. The winds, which often cause a difference in level of many feet; strong westerly winds causing a rise at one place, and easterly winds at another. These changes are irregular and transient, but often considerable in amount, ranging from two to five feet.
3. Annual variation attendant upon the seasons and confined to the year. This kind of fluctuation is a winter and summer movement. The supply from streams and rains being wholly or partially checked in the cold season, the water is gradually drawn away, lowering the general level, which reaches its lowest ebb about January or February. As spring advances, with melting snows and increased rainfall, the waters rise gradually, and attain their greatest height in June or July. They then begin to fall again to their winter level. The extreme of this variation is about 2·30 feet, and is about the same in Lake Erie as in Detroit River.
4. A rise and fall of the waters of the lakes and their connecting channels, extending through several years, and amounting to an extreme difference of five feet. Upon this kind of fluctuation Colonel Charles Whittlesey has bestowed the name of "secular variation."
The causes of this variation were long involved in much mystery. According to the old French tradition, it is independent of the seasons, and follows periodical intervals of seven years. To what extent these intervals of high and low water are regular in their recurrence, and how far they are connected with meteorological or astronomical causes, can be determined only after continuous and exact observations for a long series of years.
It is hardly more than a decade since the United States Signal Service has given scientific exactness to observations, and not over thirty years since thoroughly reliable statistics have been tabulated. Records of independent observers often differ widely, and though the writer has culled from different sources data sufficient to enable him to construct a diagram for this region, covering the past fifty years, and even more, many of these data are of uncertain value. For a period of thirty-three years, beginning with 1853, a record has been kept by the Detroit Water Board of the daily fluctuations in the level of the river, and partial records exist of other years since 1835.
In a comparison between the height of water in the river and the rainfall at Detroit, no conclusions drawn from these data will apply rigidly to the lakes above and below. The river-levels are influenced not alone by the precipitation on its borders, but by the supply from above. Other causes contribute to its irregularities—local rains, confined channel, rapid current. While a sudden increase in the precipitation will affect the broad surfaces of the lakes uniformly, a rise would take place at such times in the confined straits to a dispropor-
