454 Mineral Nutrition of Plants 



In order to determine the relationship of moisture to growth, a new 

 study was undertaken beginning in 1943 and continuing through nine 

 crops (four plant and five ratoons) until 1947. The four plant crops 

 were started at intervals of three months beginning in April, 1943. It is 

 desirable to do this if one is to use the method of multiple regression 

 in evaluating the pertinent factors in the study. If fortuitous relation- 

 ships are to be avoided, it is desirable in the collection of data to avoid 

 fortuity of circumstances. In this work, eleven samples made up of 

 individual cane tops were collected twice weekly from each of the con- 

 current crops. Twelve pilot plants were measured for growth elonga- 

 tion in each crop. At the end of each five-week period the total growth 

 of these pilots was adjusted to the total measured from forty other pilots 

 in each crop. Soil borings were made twice weekly in triplicate from 

 each crop, one boring extended down to eighteen inches and a second 

 another eighteen inches downward. Soil moisture determinations were 

 made from these. Toward the end of these studies, tensiometers (9) 

 became available and these were included. 



Complete weather data were taken from the Hawaiian Sugar Planters 

 Association weather station within a few yards of the plots on which 

 the cane was grown; the data included maximum and minimum tem- 

 peratures, sunlight measurements as gram calories per square centi- 

 meter per day, wind velocities, and humidity records. 



Since these crops were being controlled by their own crop logs, regular 

 five-stalk samples were taken at five-week intervals from each crop be- 

 ginning as soon as the plants were large enough to be sampled and con- 

 tinuing until harvest time. The fertilization of these crops followed 

 crop-log practices (j). 



After all these data were assembled, they were subjected to statistical 

 analysis using the methods of multiple regression. The first set of par- 

 tial regressions offered in Table I deals with the several factors as they 

 vary with the growth units. The growth unit is defined as the daily 

 increment increase in the volume of cane produced in a field with a 

 full stand. It is approximated by multiplying the daily elongation rate 

 by the green weight per stalk of the sheath sample. The resulting value, 

 the growth unit, correlates very well with the actual volume of cane 

 laid down. The regression of this value on the actual volume has a 

 "t" value of 17.98 and a correlation coefficient of 0.954. 



