the reading and before the final setting of the instrument for the 

 next week's run. The best way to obtain readings is to weigh 

 the entire instrument at the beginning and end of each weekly- 

 run. The weekly procedure is consequently as follows: weigh 

 the instrument, scrub the sphere, wipe off what water clings to 

 tube, stoppers and bottle, add water to bottle more than suffi- 

 cient for the next run, and reweigh. The loss in weight by evapo- 

 ration for each week constitutes the weekly reading, and five of 

 these data for each type of instrument are needed for the five- 

 week period of the seedling phase. These data will give evidence- 

 of the kind of evaporation and sunshine conditions to which the 

 plants have been subjected. 



At the end of a given series the spheres are to be dismounted 

 and dried, wrapped in paper, and sent (in a strong container and 

 by parcel post) to the Laboratory of Plant Physiology, Johns 

 Hopkins University, Homewood, Baltimore, Md. They will be re- 

 standardized, and the new coefficients will be sent to cooperators. 

 The spheres will be returned, or other ones will be sent in their 

 place (in case they seem injured) . (On operation of porous-cup 

 atmometers see: — Livingston, B. E., Atmometry and the porous- 

 cup atmometer. Plant World 18: 21-30, 51-74, 95-111, 143-149. 

 1915. Reprints may be procured from the Plant World, Tucson, 

 Ariz. The second pair of spheres is to be used during the time 

 required for restandardization. 



(4) Sunshine records are to be obtained for the period of each 

 experiment series, from the nearest U. S. Weather Bureau Sta- 

 tion operating a Marvin sunshine-recorder, these records being 

 in the form of the daily duration (hours) of sunshine, as shown 

 by that instrument. 



THE CULTURE SOLUTIONS FOR ALL PHASES. 



Introduction. 



It is obvious that there are actually an infinite number of 

 different solutions that must needs be tested if we are to find out 

 just what solution is the very best for a given plant and for a 

 given growth phase. Of course our experimentation must merely 

 approach finding this best solution and it must proceed by sam- 

 pling the range of solution possibilities (or promising portions of 

 that range) , as it were. It is highly desirable that the sample 

 solutions be selected, for the beginning, at uniform intervals 

 throughout any promising range of possibilities, and that the 

 intervals be not so broad as to let the best solution in the region 

 (for which we are looking) escape being fairly represented by 



22 



