Some powders produce more uniform pressure curves (and velocity) with less load density than others with more density. Best example was tests for a new bullet a few of us worked up loads for. A ball powder had highest load density, lowest velocity and pressure spreads, yet worst accuracy at long range. Another powder had less load density, 3/10ths grain charge spread but best accuracy.
SAAMI has a procedure that helps test data for a load be more repeatable; it's called the SAAMI Twist.
A. Cartridges to be tested should be placed in a vertical position with primer-end down in a recessed holding block.
B. When the appropriate test barrel has been properly serviced and the chronograph reset, a cartridge should be lifted vertically from the block. It should be rotated slowly, end over end, in a vertical plane through 360° pausing momentarily when the powder is at the bullet end and again when the powder is at the primer end.
C. The cartridge is then rotated slowly, a minimum amount to enter the chamber, keeping the primer end in the lowest possible position until inserted gently and carefully into the chamber.
D. The cartridge should be seated in the chamber as far as practicable with the fingers. The bolt or breech mechanism should be closed gently in order not to disturb the position of the powder in the cartridge case. The object of this method of handling cartridges is to position the propellant powder at the primer end of the cartridge case by permitting it to fall gently against the primer while rotating the case.
E. The rate of fire should not be rapid enough to cause excessive heating of the barrel. The time between rounds depends on the equipment, as the barrel may be cooled by a constant stream of air on the outside or by directing air through the bore after each ten rounds.
F. Ammunition conditioning should be between 60° - 80°F (15.6° - 26.7°C).
G. A minimum of one and up to three warming shots should be fired before firing each series for record. The velocity and/or pressure of these shots may be recorded, but should not be included in the record of the sample.