PERMIAN GLACIATION 581 



convection, evaporation, and reflection exceeded the gain due to the 

 blanketing effect of increased cloudiness. This can be tested either 

 mathematically or empirically. Both methods meet with many compli- 

 cations. Doubtless the mathematical method will in the end prove the 

 most reliable, but with our present scanty knowledge it requires a great 

 number of assumptions which expose it to constant error. Moreover, the 

 process of calculation is so intricate that the layman is almost sure to 

 make mistakes. Therefore it seems wiser to adhere to the method which 

 prevails throughout this paper — that is, let us discover what is happening 

 today and then ascertain what would have happened if the same thing 

 had occurred on a larger scale in the past. 



THE BLANKETING EFFECT OF CLOUDS AT PRESENT 



In order to find the present relation of clouds to temperature, I have 

 made use of four charts in Bartholomew's Meteorological Atlas. Two 

 show the mean cloudiness of the world in January and July. The cloudi- 

 ness is expressed in tenths of the sky which are obscured. The other two 

 charts are those of isanomalies of temperature in January and July. 

 These indicate the extent to which the temperature departs from what 

 would naturally be expected in any given place at the particular time in 

 question. For instance, if there were no winds and ocean currents, the 

 west coast of Ireland would be as cold as the east coast of Labrador. As 

 a matter of fact, in January Ireland is more than 30° F. warmer than 

 would be expected in its latitude, while the east coast of Labrador is 10° 

 F. colder than would be expected. These figures constitute the anomalies 

 for the respective regions during January. The parts of the earth more 

 tlian 30° from the equator are subject to extreme anomalies, partly be- 

 cause of the great size of the continents and partly because of the ocean 

 currents. Therefore the present investigation has been restricted to the 

 area within 30° of the equator. This is the more appropriate, since, ac- 

 cording to the assumptions of our hypothesis, it is chiefly there that 

 storminess appears to have increased during Permian times. 



In the charts of cloudiness and temperature anomalies use has been 

 made of the 126 points where the equator and the parallels of 10°, 20°, 

 and 30° north and south intersect the meridians of 0°, 20°, and 40°, and 

 so on at intervals of 20° around the world. For each point the anomaly 

 and the cloudiness in January and July have been taken, and the figures 

 have been combined so as to show whether increased cloudiness is ac- 

 companied by a rise or fall of temperature. Three different methods 

 have been p^irsued. In the first place all the points have been divided 

 into groups on the basis of their anomaly. The first group, A in Table 7, 



