Why Short Intervals Keep Appearing in Plans: The Format Is the Template

Short intervals show up in every indoor rowing plan from first-timer to finalist, and the usual explanation is that they are special. They are not special. They are a container, and containers are easy to hand to someone: one number to write down, one number to check against next week.

That is a perfectly good reason for them to recur, and it is not the reason most writing gives. This article keeps the practical case, which holds up well, and drops three numbers that were doing a lot of work without being in the papers they were attributed to.

What the format is actually for

Two things, and both are ordinary.

Repeatability. A piece you can do five or six times is a measurement. The same piece at the same split, logged week after week, is the only version of "are we getting fitter" that survives contact with a bad night's sleep. Compare that with one long piece: it gives you one number, it tells you less about where you are in a block, and the cost of a bad day is the whole session.

Visible drift. A short rep is long enough to settle into a target and short enough that you can still see the stroke falling apart in the last fifteen seconds. That is the diagnostic value, and it is genuine. What it is not is a claim about energy systems — the same number of watts does different things at different rates over different lengths of time, and the reason to care is effort, not biochemistry.

Neither of those depends on the specific shape. Four by 500, four by four minutes and six by thirty seconds are all repeatable and all show drift. The format is interchangeable; the session is not.

There is older evidence for holding the protocol fixed, and it is worth having because it is counter-intuitive. Velocity at VO₂max came out the same across two quite different ramp protocols — 20.7 against 20.8 km/h — while the time each athlete could hold at that velocity varied by 35% to 45% between them [4]. How you ran the test barely moved the measurement. Who you are moved it enormously. That asymmetry is the entire argument for logging the same piece repeatedly, and it is also why a target that is correct for one rower can be wrong for the next.

The shape inside the piece is a separate question from its length. A model built on the flow of energy in track cycling found the best strategy is event-specific and individual: all-out for the 1000 m time trial, and an all-out start settling into even power for the 4,000 m pursuit, with even small variations in pacing moving the finishing time substantially [5]. Four by 500 is long enough for you to feel that difference. Thirty seconds is too short to see it.

The work:rest ratio is one variable, not the principle

The most common version of the argument says the ratio is the thing: the work has to be hard enough to demand recovery, the recovery has to be long enough to deliver it, and everything else is detail. The ratio is real. It is also one of nine variables in the standard account of interval prescription — work intensity and duration, relief intensity and duration, modality, repetitions, series, and between-series recovery — and the paper that lays them out is careful to say that manipulating any one of them changes the physiological response [1].

So if you shorten the rest, you have changed one of nine things, and you have no basis for predicting what the session now costs you.

What the same source puts at the centre is time accumulated at or above roughly 90% of maximal oxygen uptake. That is the quantity that tracks whether a session did the work it was meant to do, and it is the thing to look at when you are wondering whether two very different formats are actually equivalent. A session that never gets there is a different session from one that spends ten minutes there, no matter how neatly the format matches.

Three numbers that were never in the papers

The sprint ceiling. A meta-analysis of how many sprints belong in a session is routinely summarised as four being enough, six too many and eight the limit. The paper concludes the opposite direction: the improvement in VO₂max from sprint interval training was 7.8% ± 4.0%, the effect of adding repetitions was possibly small and in the unfavourable direction (−1.2% ± 0.8% per two additional sprints), and the stated conclusion is that the benefit is not attenuated with fewer repetitions and possibly even enhanced — which is an argument for making SIT more time-efficient, not for capping it [2]. That is the finding's negation.

The fifteen-second intervals. A widely repeated line attributes to Billat's 15s–15s work the finding that short bouts around critical velocity produce VO₂max in six weeks. The study is an acute comparison of three single sessions in middle-aged runners aged 52 ± 5 years. It measured how long VO₂max could be held — 14 minutes on the two gentler protocols against 7 on the hard one — and blood lactate afterwards, 9 against 11 mmol/L. There is no six-week arm and no VO₂max adaptation, because nothing was trained [3]. What it actually demonstrates is genuinely useful and much smaller: the amplitude of the hard/easy contrast changes how much time you spend at the top without changing the protocol's validity.

The rate bands. Prescriptions of the form "26–30 spm at a four-minute pace, 36–44 spm at a thirty-second sprint" are attributed to the rowing physiology review and the interval-training review. Neither states them. The rowing review is about cardiorespiratory physiology, cerebral blood flow and the Valsalva-like manoeuvre of the stroke; the interval review is about programming variables. The bands are not wrong so much as unsourced, and they are the kind of number a reader cannot check, which is precisely the reason not to print them.

What the rowing studies actually found

Two rowing trials are usually cited to say polarized training works. Read carefully, neither says that.

The first put eighteen experienced rowers on twelve weeks of either all-low-intensity training or a mix, and found that improvements in 2,000 m performance and VO₂peak occurred independently of group — the between-group tests came back at p = 0.8 and p = 0.42 [6]. The group that trained almost entirely below threshold did gain more power at lactate threshold (23.5 ± 12.2 W against 5.1 ± 5.0 W, p = 0.013) and more power at 4 mmol/L lactate, and it attenuated the blood-lactate response to a given intensity more. That is a real and interesting difference. It is a difference in how the rower responds to a given pace, not a difference in how fast the rower is, and the study's own conclusion says performance and VO₂peak improved by the same magnitude either way.

The second is closer to the claim. Fourteen national elite male rowers trained for eleven weeks under two intensity distributions with total volume matched and the low-intensity share clamped at about 93%. The polarized group really was more polarized, by the index the authors defined. And no performance measure differed significantly between the groups — 2,000 m power changed by 1.5% either way, VO₂max did not change in either, and the only movement anywhere was a small increase in power at 4 mmol/L lactate in the pyramidal group [7].

If you want a defensible statement about distributions, it is this one: the evidence does not show a polarized distribution produces better performance than a pyramidal one in rowers, and it is an open question rather than a settled advantage.

How to build one

Start from your own reference rather than a table. In elite rowers the quantities that predict 2,000 m performance are power at VO₂max, oxygen uptake at lactate threshold, power at 4 mmol/L lactate, and maximal power — together 98% of the variance in speed, validated in eighteen rowers [8]. You will not have measured all four, but it tells you where the decision lives: at a physiological marker, not at a distance.

Then pick a length you can repeat, and set the target relative to what you can already hold. For a genuinely hard session the reference is close to your 2K pace or slightly faster, and the talk test is the wrong tool for it. At that intensity you are not meant to be able to speak a short sentence at the end of the rep, so a talk-test cutoff would tell you the prescription is too hard precisely when it is doing its job. Use perceived effort instead — rate the session on the same scale every time, and let the split tell you whether the target moved. There is no published number here because there is no evidence for one, and a split printed in a table describes an average rower rather than you.

Recovery is the variable to change first, because it is the cheapest. Shortening rest between identical reps while holding pace and holding the rep count is a clean single-variable change, and because the effort is identical in absolute terms, the session RPE — effort multiplied by duration — is directly comparable week to week [10]. If RPE climbs while the split holds, the same session is now costing more, and that is worth knowing before you add anything.

The low-volume model is worth knowing about if all-out work is not for you: eight to twelve one-minute efforts at peak power with seventy-five seconds of recovery, six sessions across two weeks, raised mitochondrial capacity and improved both a sprint and a sustained time trial [9]. Same container, different decision about how hard.

What this does not tell you

The rowing-specific interval literature is small, and most of what is cited is from cycling or general physiology. Treat a cross-sport result as a hypothesis about rowing rather than a result about it.

Short intervals are not a substitute for testing. A hard session develops determinants — power at VO₂max, maximal power, threshold markers — but the 2K is where you find out whether they moved [8]. Use the intervals for the weekly dose and the test for the calibration.

And the format will not rescue a bad target. The ergometer's own measurement error is a little over one percent of mean power over 2,000 m, so a change smaller than that is not a change.

What to do tomorrow

Pick one short format — five by 500 at a split you can hold five times — and log the split and the session RPE for four consecutive weeks. Do not change the rest. On the fifth week you will have four comparable numbers, which is more than most training logs ever produce, and it will tell you whether the target you chose was the right one.

If the fifth rep falls apart, shorten the rest before you shorten the piece. If the whole session feels too easy, move the target, not the format.