200 TABLES AND RESULTS OF THE PRECIPITATION, 



As the above numbers give the character of the annual fluctuation almost by 

 inspection, but few remarks are required respecting these secondary results. 



The rains in central Pennsylvania (group 1) partake of the combined characters 

 of Types I and IV, as do also those in West Virginia (group 2); in southeastern 

 Virginia (group 3) the rains are distributed according to Types VII and VIII ; in 

 western Florida (group 4) according to Type VII ; in Alabama (group 5) according 

 to Type VI, whicli modified gulf type is a very remarkable one, owing to its very 

 low summer maximum. The curves for central and southeastern Texas (groups 6 

 and 7) are peculiar; they rise to two maxima, in June and September (the principal 

 one), and fall to a principal minimum in January. The range is very large. The 

 distribution in Tennessee (group 8) and in Kentucky (group 9) is a reproduction 

 of Typo VI; the Ohio (group 10) and Indiana (group 13) curves approximate 

 more towards Type II, and Illinois (group 14) towards Type III. The curves for 

 southern and northern Michigan (groups II and 12) are alike, and present a pecu- 

 liar type, which may be called the lake-type, exhibiting one maximum in Septem- 

 ber, and a minimum in February, with a moderate range. The Wisconsin (group 

 15) and Iowa (group 16) curves ajiproach Type III ; also the Nebraska and Kansas 

 (group 17) curve, the secondary September maximum, however, disappeared. The 

 distribution of rain in New Mexico (groups 18 and 19) is similar to that given in 

 Type VIII, with a high maximum in August and low minima in January and May. 



By means of the preceding ratios of the annual distribution and the annual 

 amount given on the chart by hyetal curves, the normal amount of rain for every 

 month may readily be obtained for any desired locality. 



Permanencij of the Annual Fluctuation. — Tlie combinations made in the preced- 

 ing investigation imply that no sensible change has taken place in the law of the 

 annual distribution within the period of observation; that such is really the case 

 remains to be proved. The material collected which mainly refers to our own 

 times, and hardly reaches back into the past century, is evidently insufficient for a 

 full investigation, and obliges us to be satisfied, for the present, with a less com- 

 plete proof; but it can be shown that the secular change, if any, in the annual 

 distribution, must be very small. The best proof of the invariability of the annual 

 fluctuation, during the past hundred years, is furnished by the observations at 

 Charleston, S. C. between 1738-1759, and 1841-1861 ; these observations are 

 represented by the equations : — 



For the 22 years in the 18th century, 

 R = + 3.76 + 1.93 sin (0 + 239°) + 1.16 sm(2 + 20°) + 0.22 sin (3 + 130°,) 

 and for the 20 years in the present century, 



R = ^ 3.49 + 1.59 sin {6 + 243°) + 1.19 sin (2 + 14°) + 0.46 sin (3 + 147°,) 

 wliich equations give nearly identical results. That the difl"erences are within the 

 probable error of observation becomes apparent by a comparison of the modern 

 results among themselves by means of station Fort IMoultrie in Charleston Harbor, 

 where we have a 17 year series between 1842 and 1859. The expression for Fort 

 Moultrie is 



R = + 3.19 + 1.77 sin (0 -}- 247°) + 1.36 sin (2 -f 22°) -]-0.6S sin (3 -f 156°.) 



