the dark segment there appeared many glowing elongated spots all pointing toward a common circle below the horizon. These came and went with such marvelous swiftness that it was difficult to follow their forms with the eye. Still later, all signs of the aurora disappeared, except the primary arc, which had for a time at its lower edge a faint suggestion of prismatic colors. This rested motionless on the midnight heavens until about two o'clock, when it slowly faded, but before it disappeared it was replaced by a bewildering display of a rayed arc.
From September 1 to the 9th the temperature steadily fell. On the 8th the thermometer registered—43.1 C. This was the coldest spell of the year, and it was followed on the 9th and 10th by the most vivid and impressive auroral displays that we saw. The exhibit of the evening of the 10th began in quite the usual way, with a cloudless brightness in the south. Soon there appeared an arc with its ends about 10° above the ice. Under this arc there appeared three fragmental arcs. In the course of an hour the first arc nearly disappeared, leaving only a crescent strip, but under it there were bows more or less elliptical. These vied with each other, alternately brightening and fading, and
vanishing altogether or in parts, until after midnight. At about one o'clock they disappeared suddenly and in their place came, with an electric glow and swiftness, a bewitching array of ragged patches describing four arcs. One hour later these spread over the entire heavens, making a system of quivering, moving streamers, sweeping the skies and illuminating the snows with an effect perfectly bewildering.
This was the last great aurora that we saw, and it was followed by only one other, on September 20. The night at this time was so bright that the phenomenon was barely visible, but its form was different from any which we had seen. There were two horizontal bands; between these was an imperfect arc, and on both sides were crescent shaped patches. The whole display came and went within an hour.
It is a curious fact that the auroras usually appeared about the 20th day of the month. During the long polar night, when the boreal display is at its best, we saw very few exhibits. The phenomenon had little or no effect upon the compass, but it seemed to have some connection with the storms, for it was invariably either preceded or succeeded by violent atmospheric agitation. We did not hear any sounds,