| Abstract: | Abstract Soil tests should produce data useful for making accurate lime and fertilizer recommendations aimed at optimizing chemical conditions for plant growth or providing available nutrient levels sufficient for maximum yields. In order to attain these objectives, we believe, quicktest‐computed rates have to be made equivalent to actual values through a regression adjustment. Although it may not be most desirable to use this adjustment for each new population of soils, it may be worse to try to use a function derived from one group of soils on a second group not well represented by the first. |
