THE ERUPTION OF PELEE. 221 



When the cloud spread over Fort de France these necessarily rolled 

 out into long billows and irregular kidney-shaped forms, the lower 

 surface of which might sometimes be seen when illumined by the light- 

 ning flashes. Discharges from node to node, with the 'end on' view, 

 would be small points of light, like the sparks between the electrodes of 

 an induction machine when the two are pushed together. They would 

 also be very numerous and incessant as the individual nodes were 

 numerous, small and possessed of intensest vertical velocities over the 

 direct discharge from the volcano's throat. As the mushroom spread 

 to a distance the individual dust eddies became larger and more 

 diffuse, and the discharge between them would become less frequent 

 and more irregular, giving rise to the worm stage. With further 

 increase and more massive movements within the cloud, the same con- 

 tinually growing in volume and rising to greater height as well 

 as spreading across the line of vision of the observer below, the 

 lightning flashes decreased in frequency, and increased in length, in- 

 tensity and irregularity, producing finally the long orange-colored 

 serpents described below — the color being due to the dust layer which 

 partially obscured the dazzling flashes. 



Whatever the cause, the lightning clearly showed a gradual change 

 in character; from incessant, shifting, scintillating points it changed 

 to less frequent worm-like lines and short forking discharges, the 

 divergence being at first downward toward the crater. One flash was 

 noticed in a direction N. 30° W. having a distinct short S shape with 

 three inflections, the oq lying horizontal. At no time were any light- 

 ning flashes seen to pass from earth to cloud or vice versa; all of these 

 discharges were across the sky. At 9 :30 the continuous grumbling 

 from the volcano died away. 



At 9:43 the middle of the dust arc southward had the bearing 

 S. 20° E. This point corresponded to what at the start was the 

 zenith point of the dust balloon. 



At 9 :45 very brilliant sheet lightning occurred showing no definite 

 points or lines, but illumining the whole vault in dazzling flashes. 

 The lightning seemed to be more concentrated on the periphery of the 

 cloud. Meanwhile white cumulus clouds lifted towering thunder 

 heads to the southeast and these marked the beginning of local 

 thunder-storms over all the high points of the island. The high tops 

 of these clouds lighted by the sheet lightning showed their billowy 

 profiles white against the dark vault of dust above; some of them 

 appeared to rise to a great height, but their summits were far below 

 the black dome of volcanic dust, affording an excellent measure of the 

 stupendous height of the volcanic cloud and its electrical phenomena 

 — ^both entirely distinct from the rain-bearing thunder-storms. 



At 9 :52, looking southwest, the thin edge of the dust cloud against 



