The =barograph= (Fig. 13) is very similar to the thermograph in general appearance. The essential portion of this instrument consists of a series of six or eight hollow shells of corrugated metal screwed one over the other in a vertical column. These shells are exhausted of air, and form, in reality, an aneroid barometer which is six or eight times as sensitive as the ordinary single-chamber aneroid. The springs for distending the shells are inside. The base of the column being fixed, the upper end rises and falls with the variations in pressure. The movements of the shells are magnified by being carried through a series of levers, and, as in the thermograph, the motion is finally given to a pen at the end of the long lever. The compensation for temperature is the same as in the ordinary aneroid. A small quantity of air is left in one of the shells to counteract, by its own expansion at increased temperature, the tendency of the barometer to register too low on account of the weakening of the springs. The barograph may be placed upon a shelf in the schoolroom, where it can remain free from disturbance, and yet where the record may be clearly seen. The general care of the barograph is the same as that of the thermograph. Brief instructions concerning the care and adjustments of these instruments are sent out by the makers with each instrument. Frequent comparison with a mercurial barometer is necessary, the adjustment of the barograph being made by turning a screw, underneath the column of shells, on the lower side of the wooden case. [Illustration: FIG. 13.] Barograph records are fully as interesting as those made by the thermograph. The week’s record traced on the writer’s barograph during a winter voyage from Punta Arenas, Strait of Magellan, to Corral, Chile, Aug. 2-9, 1897, gives a striking picture of the rapid and marked changes of pressure during seven days in the South Pacific Ocean (Fig. 14). [Illustration: FIG. 14.] The following figure (Fig. 15) presents samples of barograph curves traced at Harvard College Observatory, Cambridge, Mass., during Feb. 22-28, 1887, and May 17-23, 1887. The February curve illustrates well the large and irregular fluctuations in pressure, characteristic of our winter months; while the May curve shows clearly the more even quality of the pressure changes in our summer.