The trial was designed to be boring, and that was the point.
Four greenhouses stood in a row, built to the same plan and planted on the same morning with the same variety of tomato. They got the same water, the same soil, the same warmth. One thing, and only one thing, was allowed to differ: the hours of extra artificial light given each evening. The first house got none, the second two hours, the third four and the fourth six.
The harvest, in kilograms for each square metre of floor, came out at eighteen, twenty-two, twenty-six and twenty-seven.
Read that row slowly, because it says two separate things. The first three steps each added four kilograms, which is a strong and steady return. The last step added one. Somewhere between four extra hours and six, the plants stopped being able to use the light, and the grower is now paying for electricity that the crop cannot spend.
None of that could have been said if the trial had been run any other way. Had the fourth house also been given richer soil, its extra kilogram would have belonged to nobody — light or soil, with no way to tell them apart. Holding everything steady and moving one quantity across a range is the oldest reliable trick in experimental work, and it is exactly what a slider does on a screen: one number moves, the picture answers, and the answer belongs to that number alone.