Tonnage counts the reps that cost you nothing

Sets times reps times kilos produces a tidy weekly figure that treats the easiest repetition as equal to the hardest.

Multiply sets by reps by kilos and you have a figure that looks like work done. It is not work done. It is arithmetic that happens to have kilos in it.

The ninth repetition of a set of ten is a different physiological event from the first, and everyone standing near the bar knows this. Tonnage does not know it. Ten reps at sixty kilos and six reps at a hundred can land within a few hundred kilograms of each other on the weekly total, and a coach reading that total would conclude the two weeks were roughly equivalent. They were not remotely equivalent. One of them accumulated fatigue in slow grinding sets near failure; the other moved a lot of light bars quickly and finished fresh.

What makes this worse is that tonnage is easy to inflate in exactly the direction that feels productive. Add a warm-up set, log it, and the weekly number rises. Add an extra light back-off set because the athlete is tired, and the number rises again. The metric goes up on the days when the training got softer. I have seen weekly volume charts trend beautifully upward across a block in which the actual heavy work quietly shrank, because the assistance exercises expanded to fill the time.

There is a deeper problem in the units. A kilogram lifted in a squat and a kilogram lifted in a lateral raise are not the same kilogram in any sense that matters to a body, but tonnage adds them without hesitating. An athlete whose programme includes a lot of small-muscle work will show a lower total than one doing the same amount of hard leg work, and neither number is telling you who trained harder.

The fix is not exotic. Count only sets taken within a few repetitions of failure, per movement pattern, and let the load live inside that count rather than being multiplied by it. Hard sets are the thing that drives adaptation, so hard sets are the thing worth counting. It is a cruder number and it will look less impressive in a report.

That is the point. A metric you cannot pad by adding warm-ups is worth more than a metric that rewards you for them, even when the padded one has three more digits and fills the graph better.