30 



MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 



42. We see from this table tliat tluring the months from June to September, at both of these 

 statiou.s, in two thirds of the cases, the cyclone was followed by a wind from some point between 

 ESE. and south, and this wind generally lasted more than twenty four hours. In the remaining 

 cases, the wind which followed the cyclone was from the SSW. or SW., and in these cases the 

 center of the cyclone j)assed on the east side of the given station. In the month of October (when 

 the prevalent wind is from the northeast) each cyclone was followed bj' a wind from some point 

 between south and southwest, and this southerly wind lasted more than twenty-four hours. It 

 aj)pears, then, that in the China Sea cyclones are generally succeeded by a southerly wind of con- 

 siderable duration, even in those months in which the average wind is northerly. 



From Mr. Elliott's investigations of cyclones in the Bay of Bengal, particularly the cyclone of 

 October S-19, 1882, it appears that these southerly winds, which prevail on the south side of a 

 cyclone, extend down to the equator as strong winds, accompanied by severe squalls and rain. 

 During a tropical cyclone these southerly winds appear to push forward with greater persistence 

 than the northerly winds, and this seems to be at least a part of the reason for the northerly 

 progress of the cyclone. 



These results accord substantially with those found in article 33 for West India cyclones, ami 

 show that the average direction of the progress of cyclones does not coincide with the average 

 direction of the wind for the same season of the year, but corresjionds more nearly to that of the 

 principal wind prevailing at the time of the cyclone. It is not, however, claimed that there is an 

 exact agreement between these two directions. 



43. An examination of Plate XI shows that in the middle latitudes of the northern hemisphere 

 there is a remarkable correspondence between the average direction of the progress of storm centers 

 and the average direction of the wind as shown on CoflQii's wind charts. I have endeavored to 



^ascertain whether this correspondence is exact, or whether there is a constant difference between 

 these two directions. I first made a comparison of these two directions for tlie Atlantic Ocean. 



44. In order to determine the average direction of progress of storm centers across the Atlantic 

 Ocean I measured with a protractor the bearing of the storm tracks delineated on the United States 

 International Charts. These bearings were measured for six points, viz, at the intersection of the 

 storm tracks with the meridians of 10°, 2(1°, 30°, 40°, 50°, and 00° west of Greenwich, and the 

 measurements included the observations of four years, viz, 1878-1881. Table XII shows the 

 average results of these measurements for each month of the year and for each of the six points 

 above mentioned. The latitudes named at the top of the table are the average latitudes corre- 

 sponding to the given directions : 



Table XII. — Direction of storm tracls. 



45. I next endeavored to determine the average direction of the wind for the entire year at 

 several poiiits on the Atlantic Ocean, as near as possible to the points corre^pouding to the pre- 

 ceding measurements. Table XIII is derived from Maury's Pilot Charts of the Atlantic Ocean, 

 and shows the number of wind observations for sixteen points of the compass, for each of the 



