should be taken to have the clock keep good time. These adjustments can readily be made by means of a screw and a regulator, respectively. The thermograph should be exposed in the instrument shelter with the other thermometers. The sheets should be changed, the clock wound, and the pen filled once a week, preferably on Monday, at 8 A.M., or at noon. The continuous records written by a thermograph are a valuable addition to the fragmentary observations which are the result of eye readings of the ordinary thermometer. From the former any omitted thermometer readings may be supplied. The interest of thermograph records may be seen in the following figure (Fig. 12), in which curves traced under different conditions are reproduced. Curve _a_ illustrates a period of clear warming weather at Nashua, N. H., April 27-30, 1889. Curve _b_ was traced during a spell of cloudy weather at Nashua, accompanying the passage of a West India hurricane, Sept. 13-16, 1889. Curve _c_ illustrates the change from a time of moderate winter weather to a cold spell (Nashua, Feb. 22-25, 1889). Curve _d_ exhibits a steady fall of temperature from the night of one day over the next noon to the following night, during the approach of a winter cold spell (Nashua, Jan. 19-21, 1889). Curve _e_ shows a reverse condition, viz., a continuous rise of temperature through a night from noon to noon (Nashua, Dec. 16-17, 1888). Curve _f_ shows the occurrence of a high temperature at night, caused by warm southerly winds, followed by cold westerly winds (Cambridge, Mass., Nov. 30-Dec. 1, 1890). Curve _g_ illustrates the sudden rise of temperature due to the coming of a hot, dry wind (chinook) at Fort Assiniboine, Mont. (Jan. 19, 1892). A study of such records leads to the discovery of many important facts, which would be completely lost sight of without a continuous record. [Illustration: FIG. 12.]