26 THK ASIATIC FUE-SEAL ISLANDS. 



side of the mountains. In deep shadowy gullies it often remains all summer, and in 

 cold seasons, as for instance 1895, large drifts still remain unmelted as late as 

 September, even at the level of the sea. 



Drift ice seems to be of rare occurrence in recent times. I do not know how much 

 reliance can be placed in old Pitr Burdukovski's story to me that formerly, say about 

 1850, " drift ice was yearly observed coming from the north in large masses." Certain 

 it is that Steller expressly states that during the winter no ice collected in the sea. 

 (Ber. Ins., p. 270.) 



To complete the meteorologic account I may mention that thunderstorms are of 

 rare occurrence on the Commander Islands. In 1879, on November 19, Mr. Krebs, 

 after a residence of eight years in the main village on Copper Island, expsrienced the 

 first thunderstorm. In 1881^ on February 8, he records " a stroke of lightning and a 

 short, but strong thunderclap about 7 p. m." Mr. Chernick, in Mkolski, Bering 

 Island, reports "thunder and lightning" on September 12, 1878. I myself observed 

 a thunderstorm passing over Nikolski, September 18, 1882. The first lightning was 

 observed at 9^ 58" p. m., local time; wind SW., 13 miles an hour; barometer, 29.552 

 inches; temperature of air, 52.2° F. ; clouds, cumulo- stratus, 8, direction SW.; intervals 

 between first lightning and thunder, 96 seconds; sixth thunderclap (lOi^ 25"* p. m.), 

 12 seconds after lightning; tenth, 40 seconds; eleventh lightning before thunder of 

 tenth. This was the last distinct thunder heard, 10" 35™ p. m. After that continued 

 distant lightning lit up a narrow strip along the northern horizon. ISTo lightning seen 

 after 11" 10™ p. m. 



Aurora borealis is equally scarce. At Nikolski, on K"ovember 15, 1882, 1 observed 

 a faint northern light at 12" 30"" a. m., local time, extending to about v 'Urs. majoris. 

 On November 17, 1882, 1 observed another at 10^ 40™ p. m., local time, consisting of 

 a uniform greenish white, light below, above which most of the time a large rosy space 

 was seen filling the arch between y and 77 JJrs. majoris; a similarly colored but often 

 broken arch extended through the constellations of Gygnus, Cassiopeia, Gemini, and 

 Auriga, sometimes fainter, sometimes more fiery, especially in Gygnus. Very seldom- 

 the red color filled the space between the rosy spot below Ursa major and the upper 

 arch, and then only for a few seconds. At 11 p. m. the sky became so overcast as to 

 cut off further observation. 



Corresponding observations made at St. Paul Island, Pribilof group, from 1872 to 

 1883, and published by the United States Weather Bureau (Fur Seal Arb., 11, App. 

 pp. 591-593), afford means of exact comparison between the Eussian and the American 

 seal islands, except as regards mean temperature, the latter being obtained on St. Paul' 

 from observations made at 7 a. m., 2 p. m., and 9 p. m. 



But even a comparison of the mean temperature affords several very interesting 

 results. Thus, while the annual means apparently differ but slightly, there is also the 

 same relative proportion between the various months from December to September. 

 But while the figures representing the mean temperatures for these months are higher 

 on Bering Island than on St. Paul, those of October and November are higher on the 

 latter. The chief exception from the relative proportion between the months is shown 

 by the mean temperatures of August, which is about 4 degrees higher than July and 

 September in Bering Island, but only about 2 degrees in St. Paul. 



Turning now to the maximum temperature, it will be seen to be 63° F. in Bering 

 Island as against 62° on St. Paul. But on the other hand, while the minimum 



