ATMOSPHERIC DISTURBANCES. These appear mainly in winds of great violence, frequently of tornadie intensity, which pass in advance of the clouds of eruption (in this sense compar able with the wind that is forced ahead of a snow avalanche), and others whose course is di rected toward the volcano, and which are seem ingly indraughts into vacual areas made by the sudden explosion of steam. These so-called con trary currents have been noted particularly in the eruptions of Tarawera. in IS86, of Bandai San, Japan, in 1SSS, and of Pelee. They are also at times very destructive, almost as much so as the direct winds which are scattered about by the outgoing blasts. It is not unlikely that most of the destruction that followed the explosion of Bandai-San, in 1888, and which resulted in the annihilation of the greater part of seven vil lages, was the work of this form of wind. Many attempts have been made to associate volcanic outbursts with particular phases of the moon and sun (sun-spot periods) and with special meteoro logical conditions of our atmosphere: and on these assumed relations special prophecies as to future volcanic happenings have been built. But
no relation of this kind has so far been found, nor is there well-founded reason for believing that it in any way exists. Many of the most violent eruptions have taken place during storm periods: others of equal violence at times of almost absolute atmospheric stability. For at least six days preceding the May eruption of Pelee the barometric variation at noon was only one millimeter, the barometer marking steadily at this time 702 millimeters. This condition also existed at and before the time of Pelee's sec ond death-dealing eruption (August 30. 1902), and it was noted in the great eruption of Vesuvius in 1794.