thermometers being made of unsuitable materials. When any wood but well-seasoned boxwood is used, it soon warps on exposure to the weather, and the tube breaking, the thermometer is rendered useless. Porcelain scales also occasionally go to pieces, and the tubes are at the same time frequently snapped. The best frameworks are those made of zinc.

98. Box for Thermometers.—In order to ascertain the temperature of the air, it is necessary that thermometers be protected from the direct and reflected rays of the sun, and at the same time have the benefit of a free circulation of air.

No arrangement can completely fulfil both these conditions. For if they be completely protected from solar radiation, the circulation of the air must be unduly interfered with ; and if the circulation of the air is quite unimpeded, the thermometers are unduly exposed to radiation. All, therefore, that can be secured is a fair compromise between protection and circulation. The best and cheapest contrivance yet devised to meet these requirements is

there

101,

Fig. 13. the louvre-boarded box for thermometers, constructed by Thomas Stevenson, C.E., Edinburgh, and now extensively used by the observers of the Scottish Meteorological Society, and other meteorologists. A figure of the box, fig. 13, is here given, with the

102.

door let down to show the hanging of the thermometers inside. Fig. 14 shows the simple and ingenious method by which the louvre-boards are fixed.

99. It is screwed to four posts firmly fixed in the ground, and these posts and the box itself are painted white, being the colour which absorbs least of the sun's heat. The posts are of such a length that when the minimum thermometer is hung in its place it is exactly four feet from the ground. This height is an essential point in the arrangements of the observatory, owing to the very great differences which frequently obtain between the temperature of the air at four feet, at the surface of the

Fig. 14. earth, and at intermediate points.

100. Placing of the Thermometer-Box. — The box should be placed at some distance from walls, or other objects likely to be heated by the sun ; in an open space; and over old grass to which the sun has free access during the greater part of the day. For if it is placed on the north side of walls or buildings, the thermometers do not indicate a sufficiently high day temperature nor a sufficiently low night temperature for the average of the district where they are placed. And if it is placed over black soil, which is more highly heated during day, and cooled to a greater degree during night than grass, the maximum temperature will be too high, and the minimum too low.

101. Mean Daily Temperature.-If the thermometer be observed once each hour, or twenty-four times a-day, and the sum of the observations be divided by 24, we shall obtain the mean temperature of that day. Observations of this nature, extending over considerable periods, have been made at Greenwich, Leith, Rome, Madras, and other places in different parts of the world, and the hourly means for each month have been calculated and published. These tables show that there are two times in the day when the temperature is the mean, occurring generally in winter between 9 and 10 in the morning, and between 9 and 10 in the evening, and in suinmer about an hour earlier. Hence 9 A.M. and 9 P.M. are the hours best suited for observing when only two observations are made in the day. The means of the observations at the following hours differ very little from the true mean temperature-4 A.M., 10 A.M., 4 P.M., and 10 P.M.; 6 A.M., 2 P.M., and 10 P.M.; 7 A.M., noon, and 10 P.M.; and generally the mean of four hours at equal intervals will be found to give a result not far from

the true mean.

102. When three observations are made daily, the best hours are

9 A.M., 3 P.M., and 9 P.M. The observation at 3 P.M., being near the time when the temperature is highest, is of great value in reference to the climate of the locality, as well as to other considerations of more strictly scientific interest.

103. It might have been supposed that the daily minimum would have occurred at the rising of the sun, or just before its rays had begun to heat the air; but observation is unanimous in showing that the minimum happens some time before the sun rises. Within the tropics, and in temperate regions during summer, it occurs about half an hour before sunrise ; but in temperate regions during winter generally two hours, and at some places as much as three hours, before sunrise. As causes concerned in raising the temperature of the air before the sun's rays begin directly to influence it, may be mentioned—(1) heat reflected from the upper regions of the air, already heated as well as lighted up by the sun before it appears above the horizon ; (2) heat liberated from the deposition of dew during the night; and (3) the slight influx of air from the warmer east towards the place of greatest cold, as evidenced by the daily fluctuations of the barometer, perhaps contributes in a small degree to the same result.

104. Within the tropics, and in temperate regions during winter -in other words, where the daily range of temperature is smallthe maximum temperature occurs about an hour and a half after the sun has passed the meridian at noon. But in temperate climates during summer the maximum does not take place till from 2} P.M. to 3} P.M.

105. To this general law of the time of its occurrence there are two interesting exceptions. At many places in tropical countries it occurs about noon, or sometimes a little earlier. Thus the temperature at the coast is lowered by the sea-breeze, which begins to blow before noon, and during the rainy season the temperature at this hour is still further lowered by the clouds which overspread the sky. At the Hospice of the Great St Bernard, 8174 feet above the sea, the daily maximum temperature occurs within an hour after twelve o'clock. The air being comparatively rare at this great height, the effects of solar and terrestrial radiation are more immediately felt than at lower levels. Hence, though the heat received by bodies exposed to the direct rays of the sun at noon is very great, yet the dispersion of the heat by terrestrial radiation is so rapid that in an hour after the sun has passed the meridian, the temperature of the air begins to fall. At the Great St Bernard, at 3} P.M., the temperature is 10.0 lower than it is about noon, whereas at lower situations it is from 10.5 to 20.0 higher.

3

10

106. In Melville Island, in the Arctic Ocean, there is no trace of a daily period during the absence of the sun in December and January; at Fort Bowen, the highest occurs in December at 9 A.M., and the lowest at 7 P.M., but at this place in January there is no trace of a daily period; at Igloolik, in North America, there is a daily fluctuation, but it is indistinctly marked. On the other hand, at Matotschin-Schar, in Spitzbergen, the daily range of temperature is well marked—the maximum during December occurring at 9 P.m., and the minimum at 8 A.M.; and during January, the maximum at 3 A.M., and two minima at 9 A.M. and 9 P.M.; or taking the mean of the whole period when the sun is invisible at this place, the maximum occurs at 8 P.m., and the minimum at 8 A.M.

107. Mean l'emperature deduced from Maximum and Minimum Temperatures.—Of late years, since the invention of selfregistering thermometers, the mean temperature has been more commonly deduced from observations of the highest and lowest daily temperatures. How far does the mean of these two represent the mean temperature ? To answer this question, I have compared the mean temperature deduced from hourly observations with the mean temperature of the daily extremes at 27 places in different parts of the globe. On an average of six months of the year, the difference between the two does not exceed the third of a degree ; the difference for any month seldom exceeds a degree, and the mean annual difference seldom more than half a degree. At Rio Janeiro the difference for any month does not amount to. 00.3, whereas at Catherinenburg, in the Ural Mountains, it exceeds this amount in every month but one. In some places the differences are all in excess, in others they are all in defect; in some places an excess occurs in winter and a defect in summer, and in other places vice versa. In most places the great difference is in October and November, but in a few places the reverse holds good.

108. Comparing the Leith and Greenwich observations, we find that the mean annual deviation at Leith is 0o.2, at Greenwich, 09.7; the lowest monthly deviation at Leith, 09.1, and at Greenwich, 09.2; and the highest at Leith, 0°.6, and at Greenwich, 19.1. It would appear that these differences are dependent almost entirely on local peculiarities, such as the open or confined situation of the thermometers, the covering of the ground over which they are placed, the protection from radiation, and the degree of freeness with which the air circulates around them.

109. At the stations in connection with the Scottish Meteorological Society, the mean of the daily extremes is accepted as the

alte

;

mean temperature, and most meteorologists now adopt the same mean. Others, however, apply to this mean a correction in order to bring it to what is conceived to be the true mean temperature. This is a practice which for many reasons ought to be discouraged. The mean of the daily extremes gives observation alone; whereas the corrected” means are vitiated by misleading hypotheses. If an observer in the north of England gives the corrected" mean temperature of August 1865 as 57o.3, what does this mean? If he has adopted the Greenwich correction for that month, the observed mean was 58o.2, but if the Leith correction it was 57°.4. If two observers near each other should use the one the Leith correction and the other the Greenwich, a difference in the climates of the two places would be indicated which has in reality no existence.

110. Palue of Extreme Temperatures.—Self-registering thermometers are of great practical value in recording the extreme temperatures, which in reference to their effects on health and vegetation must be regarded as most important elements of climate. The subject of extreme temperatures is one of paramount importance, especially where the transitions of temperature are sudden and violent. In the north-western parts of the United States of America the temperature in spring often rises to 83° during the day, and falls to freezing during the night. Under such a cliinate the vital functions of plants whose tissues abound in sap are called into activity during the day, but this sap being frozen during the night, the vessels containing it are ruptured by expansion, and the plant, if not totally destroyed, is so seriously injured that its successful cultivation becomes precarious and uncertain. The same risk is not incurred in such places as Great Britain, where the seasons shade into each other by nice and almost insensible gradations. It is, however, everywhere desirable to arrive at some definite information as to the probability of occurrence of certain extreme temperatures that may take place in the different seasons, and this can only be done by observations of extremes carefully made and recorded from day to day.

111. Importance of resolving Mean Temperature into the Extremes which compose it. -- Every mean temperature may be considered as a composite element, made up of the niean temperature of the day and the mean temperature of the night. Hence the same mean temperature often stands for two things essentially different. Thus, Madrid in the centre of Spain, and Menton on the Gulf of Genoa, had the same mean temperature of 72°.8 during September 1865. But the climates of the two places