supposed that by this arrangement a symmetrical curve of sound would be obtained; and we think this would have been the case had the wind remained constant in direction. It did remain nearly the same during the time of describing the first, second, and third courses, and only slightly varied during the fourth; but previous to running the fifth and sixth courses the wind had changed to a direction nearly at right angles to its first course.

As is shown in Fig. 7, the first, second, third, and fourth courses form a normal curve of audition; the fifth and sixth courses, however, give discordant results, being much longer than a symmetrical curve would indicate, showing a change in the condition of the medium from that which existed during the running of the other courses; this change was evidently that of the wind, which, veering, as we have seen, through an arc of a little more than 90°, brought it nearly at right angles to the fifth course, and approximately in the direction of the sixth course; the wind also increased its velocity. These changes are sufficient, without other considerations, to give a rational account of the phenomena observed. They both tend to increase the distance at which the sound would be heard.

In these experiments, as in subsequent ones, it is to be regretted that for want of balloons the motion of the air above could not be ascer. tained, as was done at Sandy Hook in September, 1874. Previous to sailing from the depot at Staten Island attempts had been made to secure a supply of toy balloons, but none could be found at that time in the city of New York. Arrangements were therefore made for procuring a reservoir of condensed hydrogen, by which India-rubber balloons could be inflated at the time they were wanted. Unfortunately this apparatus did not arrive in time to be of much avail in this series of experi. ments. Besides this, on account of the smallness of the balloons, the ascent was too slow compared to the horizontal motion to indicate the direction of the wind at a considerable elevation above the points of observation. They were however of use in pointing out definitely the direction of the wind and the changes it was undergoing. Moreover, at the time of leaving New York we were only able to procure one anemometer, whereas we ought to have had a number, one for the top of the tower, one for the bottom, and one for each vessel.

Experiments of September 3.-Barometer, 30.02 inches; thermometer, dry bulb, 720.5 F. wet bulb, 70°; wind from the east, but too slight to move the cups of the anemometer; it soon however spang up from the opposite direction, in which it continued during the remainder of the day, attaining a velocity of five and a quarter miles per hour.

In these experiments two light-house steamers were employed, the Mistletoe and Cactus, which enabled us to obtain the results in half the time, and thus to obviate the effect, in some degree, of any change in the direction of the wind. On this occasion the sound was noted at the light-house as it converged to a center from the whistle of each vessel,

and also simultaneously by each vessel as it diverged from the vertical siren.

We were enabled, in this way, to produce two curves by a reverse process. These are plotted in Fig. 8, and exhibit a remarkable degree of similarity. The corresponding parts of the two curves, being in each case reversed, exhibit the fact that, through the same space in opposite directions, the audibility of the sound was similarly increased with the wind and diminished against it. The effect however of the wind in the experiments of this day was less marked than on any in the whole series, and conséquently the two curves of audition more nearly ap proximate circles.

We can see in this result no other effect than that which would be produced from a wind flowing with a uniform but slow velocity at the surface, but having a slightly increased velocity above. Had there been no wind, according to this view the two curves would have exhibited two concentric circles.

Experiments of September 4.-Barometer, 29.85 inches, falling; ther mometer, dry bulb, 77° F. wet bulb, 730.25. Wind south by west, twelve and one-fourth miles per hour at the top of the tower and nine and one-fourth miles below; variable.

These experiments were also made with two vessels. The distances and directions are given in Fig. 9. With the exception of the fourth course of the Cactus, the other courses would form nearly a symmetrical curve, but in this case the sound of the whistle of the Cactus was lost at the point a at a distance of one mile, and was afterward regained at the point b, and continued audible until the steamer reached the point c.

This presents one of the abnormal phenomena of sound which might in part be accounted for by the existence of a flocculent cloud between a and b, but why the sound could be heard so much farther in this direction than in the others is not easy to explain on that hypothesis.

The line b c was described after all the lines of Fig. 9 had been completed, and therefore the curve given in the figure correctly represents the boundary of the area of audition while these courses were being run, the point a being the termination under that condition of the fourth course of the Cactus. To explain the abnormal line b c, we have only to suppose that a change in the velocity of the wind afterward took place, by which its opposition to the sound-wave was diminished; this will account for the greater length of the line; the change however did not reach the light-house until after the vessel had passed the point b.

As affording evidence in support of this hypothesis, it may be men. tioned that on examining the records of the Signal-Service, of which there is a station at New London, seven miles north of the position at which these observations were made, it was found that the wind in the morning of that day was south, in the afternoon southwest, and in the evening northwest, and that it was probable, as in other cases, that the wind had changed above while the part of the course b c was run.

Experiments of September 6.-Barometer, 29.93 inches; thermometer, dry bulb, 740.5 F. wet bulb, 67°; wind from northwest to southwest, seventeen miles per hour. The wind, though of higher velocity than on any other occasion, was variable. On this day the experiments were principally made with the Mistletoe. The Cactus, being obliged to leave on other duty, ran one course a distance of two-thirds of a mile before the sound of her whistle was lost at the light-house. She afterwards steamed off in the direction C b (Fig. 10), noting the sound of the siren, which was lost at the point b, afterward regained, and heard distinctly ten and one-half miles distant.

During the passage of the first course of the Mistletoe, the wind at the surface and above was from southwest, the latter being indicated

by a cloud passing the zenith. During the second course the wind was variable, changing its direction about 90°, principally from the northwest; while during the third course the wind was again from the southwest. The long course of the Cactus marked on the figure indicates the sound of the siren, from the center outward, as it was heard seven and one-fourth miles, then lost for an interval, and afterward heard again at a distance of three and one-fourth miles farther, making, in all, ten and one-half miles.

Experiments of September 7.-Barometer, 30.1 inches; thermometer, dry-bulb, 73° F. wet-bulb, 62°; wind, eight miles per hour above, and five miles per hour below, tower. The wind was variable, as indicated by the letting-off of balloons, which however did not rise to any great height. The direction of the wind is shown in Fig. 11 by arrows. There is nothing remarkable in the curve of audition of this day. It indicates, as usual, a greater distance toward the side on which the sound was moving with the wind.

Experiments of September 8.-Barometer, 30.3 inches; thermometer, dry-bulb, 710 F. wet-bulb, 640.5; wind, west-southwest, fifteen miles per hour at top of tower, nine miles per hour below. Fig. 12 indicates the curve of audition of the vertical siren as compared with that of the