THE elastic wave in. the atmosphere caused by the explosion of Krakatoa on August 27, 1883, travelled, as is well known, several times around the earth at the average velocity of about 315 metres per second. That is, at a velocity that averaged about sixteen metres per second less than that of sound in air at normal pressure and o ยฐ C. temperature. Other long-distance air waves also have seemed to travel comparatively slowly. Presumably, therefore, this phenomenon is owing to some persistent and ubiquitous cause.
The effects of wind on the velocity of " sound," whether audible or inaudible, are well known, and were more or less averaged out in the Krakatoa and other similar observations. Hence, provisionally, we may regard the phenomenon in question as pertaining to still air and the velocity as being relative to a fixed point on the surface of the earth. Conceivably, then, the comparatively slow velocity of far-off " sounds " may be apparent only and owing to a more or less zigzag path in a vertical plane, resultin.g from reflections and refractions. Or, perhaps, to travelling part of the way, at least, through the upper air, provided the velocity of sound is decidedly less at considerable heights than at the surface of the earth.
To test the adequacy of these causes (the second, though necessarily implying also some of the first, seems to be of the right order), and for whatever other purpose they may serve, I have computed the average winter and summer velocities of sound at various altitudes from sea-level to 16 kilometres, as per the closely known values of density and pressure in the free air over northwestern Europe? These velocities would, of course, in. every case, be directly proportional to the square root of the absolute temperature at the place under consideration if the * Communicated by the Author. 1W. J. Humphreys, " Physics of the Air," regular edition, P. 72.