. . "Since we can get better answers from positive controls, this type of error propagation analysis is rarely needed. 5) To some extent, you MAY be trying to add SYSTEMATIC error that might be introduced by changes salinity, light, and other factors to the experimental uncertainty (random error)." . . .