NATUKAL PniLOSOPIIY. 155 



weather and for foretellins: weather, to take them at stated and 

 r('o:iihir intervals, so that the variations at those periods may be 

 noted, and, if required, plotted out on a chart. Indeed, for ob- 

 taining quick and useful comparisons, there is nothing compared 

 to the phm of projecting the curves of atmospheric variation on 

 the charts specially prepared for that purpose ; it enables one at 

 a ghmce to see the variations of the barometer during the past 

 day, saving the bother and calculation necessary where the observa- 

 tions are simply noted down as so many figures. But there is one 

 great ol)jection attendant upon observations of this nature; how- 

 ever carefully they mr^y be recorded or described on charts, they 

 are but observations of the time only, and show nothing more. 

 For instance, the heights of the barometer at the two usual times 

 of observing, in the morning and evening, are recorded, and a 

 line drawn on the chart from the one point to the other is assumed 

 to show the variation between those times. True, it does to some 

 extent, but only to the extent of the difference of the two. In 

 stormy or unsettled weather the rise and fall of the barometer may 

 be considerable between the two periods of observation, and yet 

 it is possible that at the two periods the observed indication will 

 be precisely the same. The chart would consequently show an 

 even state of pressure, whereas the oj^posite would be really the 

 ciise. Accurate results can, therefore, only be obtained when the 

 observations are made hourly, or, at least, at very frequent inter- 

 vals. This is, as far as regards personal observation, quite im- 

 practicable for the generality of observers ; and to give a true and 

 faithful record of the variations of the barometer from minute to 

 minute and from hour to hour, we can only look to mechanical 

 means for bringing about this much-desired result. 



Among the plans suggested very few have been ever practically 

 carried out, and of those we have seen their great expense proves 

 an almost insurmountable barrier to their adoption. The baromet- 

 rograph seems to combine simplicity with cheapness, and accu- 

 racy with ease of observation. The records are continuous and 

 comparable, and are produced by the variations of the barometer 

 known as the aneroid. The pressure of the atmosphere affects 

 four metallic boxes, as in the ordinary aneroid, having their upper 

 and under faces undulated ; a vacuum is made in each of them 

 separately, and they are attaclied together in one series, so that 

 for an equivalent variation of pressure the movement is four times 

 greater than it is for one box only. A very strong flat steel spring 

 acts upon the barometric boxes in an opposite direction to the at- 

 mospheric pressure. This spring controls the indicating lever by 

 means of a connector, which receives the action from the extrem- 

 ity of the spring, and communicates it to the lever at a point very 

 close to its axis, whence it follows that a considerable multiplica- 

 tion of movements is the result. 



The indications of the movements of the lever are registered 

 in the following simple manner: A cylinder is revolved by the 

 regular movement of an ordinary pendulum time-piece ; it makes 

 a complete revolution in one week, and carries a glazed paper, 

 which has been smoked black by means of a caudle. At the ex- 



