33 



might be supposed that the average light at this time would be the 

 greatest. This is not ahva}-s true, however, as cloudy weather is 

 an important factor. 



September gave the highest average outdoor light conditions, 

 followed by August and July, although in the greenhouse the high- 

 est monthly record was obtained during July and June. It should 

 be pointed out however, that some of the highest daily records 

 were obtained in June. In these particular records, made in 1905, 

 October 5th gave the highest, but June 22d was practically as 

 good, and the days following this date were exceptional for the 

 amount and intensity of light. 



Records were also made for six months, from November to 

 April, at corresponding hours to those previously given. These 

 records were made for another purpose, under an ordinary thick 

 walled bell glass 12 inches in diameter and 20 in. high. (See p 13). 

 The bell glass projected vertically through the roof, and the re- 

 ceiving tube was suspended in the middle of the bell glass, its lo- 

 cation being such that it received no shadows; and the rays of 

 light, passing through the circular bell glass, were apparently fo- 

 cussed on it. The light conditions under the bell glass proved to 

 be quite dififerent. The records ran higher than those in the green- 

 house, while we have average monthly differences of 22 percent, 

 existing between the light in the greenhouse and that out of doors. 

 This difference amounts to only 14 percent, under the bell glass. 



The principal facts brought out by the tests are first, — the 

 type of constiiiction affects the light in greenhouses, and second. 



FIG. 5 — Showing longitudinal and cross sections of greenhouse with 

 partitions running north and south, and illustrating the effect of poor light 

 conditions. The oblique lines represent the path of the sun's rays. 



the curvilinear type appears to be superior to others in this re- 

 spect, owing to the smalle.r amount of timber at the eaves to cast 



