NO. 4 TEMPERATURE VARIATIONS IN THE NORTH ATLANTIC 169 



ment, but the times of maximum and minimum came after the cor- 

 responding times of minimum and maximum of sun spots. The 

 time interval varies from about six months to about two and a half 

 years. In the mean it was about one year eight months. He there- 

 fore concluded that several months after variations in the spotted 

 surface of the sun there follow corresponding abnormal air pressure 

 variations. He connected also the famines of India with these air 

 pressure variations by showing that times of famine follow his 

 atmospheric waves of high pressure. 



By his above mentioned spectroscopic investigations of sun spots 

 beginning with 1870, Sir Norman Lockyer in the year 1886 regarded 

 it fairly certain that the sun is warmest at sun spot maximum. At 

 sun spot minimum the widened lines in the sun spot spectrum cor- 

 responded generally with the lines of iron and some other known 

 metals, but at maximum the most widened lines were the so-called 

 unknowns, which had not been observed in the spectrum of ter- 

 restrial elements. He therefore provisionally assumed that the sun 

 was not only warmer at sun spot maximum, but warm enough to 

 dissociate the iron vapor. 



In the year 1900, Sir Norman Lockyer and William Lockyer pub- 

 lished a discussion of the observations of the most widened spectro- 

 scopic lines for a period of more than twenty years. They showed 

 that the two kinds of spectroscopic lines experienced regular and 

 opposite periodic variations at least up to the year 1894 or 1895. 

 These variations were such that when expressed graphically in 

 curves the curve for the iron lines tended upwards when the curve 

 for the unknown lines tended downwards, and vice versa. This 

 relation continued unchanged up to the year 1895. At certain inter- 

 vals the two curves must therefore cross one another and this should 

 occur according to the above mentioned hypothesis at the time when 

 the temperature of the sun had a mean value. These crossing 

 points lay as they found almost exactly in the mean between maxi- 

 mum and minimum of sun spots, that is to say, midway between 

 those times when one should assume that the sun was warmer or 

 colder than the mean. 



In discussion of the variations of the solar prominences (1902) 

 they found that the prominences on the whole varied in the same 

 way as the sun spots, but that within the eleven-year sun spot period 

 there occur three well marked shorter periods of about three and a 

 half or 3.7 years in the variations of the prominences. These three 

 periods occur so that while the maximum of the middle period 



