Isolate
Hold everything else constant and change one thing; a second change voids the result.
Change only one thing.
Seen in Josef Albers, Interaction of Color, 1963
Try it on your work
Change exactly one physical property of the thing you're working on (distance, height, brightness, order) and keep every other property fixed. Use or observe it changed for one session. Whatever differed is attributable, for once.
Say the same sentence in the same words to two different audiences, changing only the audience. The difference in reception belonged to the rooms, not the sentence; you've now measured the rooms.
Run one cycle of your process with a single parameter altered and a written prediction, made beforehand, of what will differ. Compare against the prediction, not against your impression.
How it has been used
Hold everything constant and change one thing: the experimental method itself, the A/B test, the recipe altered by one ingredient. The move forbids the second change: touch two variables and whatever you learn belongs to neither.
Where it leads
Grouped under Pull the parts apart