Negative Splits and Other Pacing Strategies — A Research-Grade Synthesis

Abstract

Pacing is the controllable variable on top of physiology in a 2K or 5K. The peer-reviewed literature shows that elite 2K ergometer competitors adopt a reverse-J pacing profile — a fast first 500 m, a slowest third 500 m, and a small finishing lift — rather than a strictly even split ([1] Garland 2005, Level 4). Well-trained rowers who have not yet familiarised with the test pace row a positive split on trial 1 and shift to a reverse-J on trials 2–3, while performance time stays inside the noise band ([2] Gee et al. 2013, Level 2b). The degressive (fast-start) strategy is the least-preferred strategy for trained rowers and produces more time at high blood lactate and high RPE than stable or progressive pacing, even though final lactate and final RPE are similar ([3] Boillet et al. 2022, Level 2b). A fast start does accelerate pulmonary VO2 kinetics when preceded by priming exercise, but the direct evidence is from cycling rather than rowing and is bounded to short severe-intensity efforts ([4] Brock et al. 2018, Level 2b; [8] Turnes et al. 2014, Level 2b). The honest read: there is no single best pace plan, but there is a small set of strategies that fit most rowers, and the practical recipe is the reverse-J or controlled-positive with a deliberate end-spurt for trained rowers, and even or slightly positive for novices ([9] Concept2 — Pacing blog, Level 5; [10] Concept2 — 2K test blog, Level 5; [12] British Rowing — Go Row Indoor tests, Level 5). The AI coach that anchors a rower's reference pace on a familiarised 2K and lets the first-trial habituation sit separately is reading pacing correctly; the coach that treats trial 1 as a calibration of physiology is not ([2] Gee et al. 2013, Level 2b; [5] Schabort et al. 1999, Level 2b; [6] Ingham et al. 2002, Level 2b).

What a pacing strategy is

A pacing strategy is the rower's plan for how to distribute effort across a fixed-distance piece — which 500 m to push, which to ease, what stroke rate to settle into, and whether to leave anything for a finish kick ([9] Concept2 — Pacing blog, Level 5; [10] Concept2 — 2K test blog, Level 5). The pace plan is the in-test mirror of the training distribution: a rower who trains mostly easy with a few hard sessions will approach a 2K with a plan that protects the back half, and a rower who trains mostly steady will approach it with a different shape ([16] Seiler & Kjerland 2006, Level 4; [17] Rosenblat et al. 2025, Level 2a; [20] Filipas et al. 2022, Level 2b). The pace plan is also the rower's only controllable source of variation on top of physiology on test day — sleep, hydration, warm-up, and damper are also controllable, but the pace plan is the one the rower writes down beforehand ([14] Concept2 — Damper setting blog, Level 5).

The standard vocabulary has four shapes:

• Even split — the same target split for every 500 m. The plan is simple and the rower can hold the same stroke rate and power output throughout. The empirical literature shows elite 2K rowers rarely use a strict even split ([1] Garland 2005, Level 4).
• Negative split — the second half is faster than the first half. This is the strategy most often recommended in older coaching literature. The peer-reviewed rowing evidence does not show it is the strategy of elite 2K competitors ([1] Garland 2005, Level 4).
• Controlled-positive (or "slightly positive") — the first 500 m is a few seconds faster than the average, the middle is at or slightly above the average, and the back half is at or slightly below. This is the empirical pattern of elite 2K rowers ([1] Garland 2005, Level 4).
• Reverse-J (or "conservative-start, end-spurt") — a fast first 500 m, a slowdown through the middle, and a finishing lift in the last 500 m. This is the pattern well-trained rowers converge on after a familiarisation trial ([2] Gee et al. 2013, Level 2b; [1] Garland 2005, Level 4).

What the elite 2K pacing profile actually looks like

The 2005 [1] Garland paper is the single most direct anchor for what elite 2K pacing actually is (Level 4). Garland screened 1,612 on-water race profiles from the 2000 Olympics and 2001–2002 World Championships; 948 met the inclusion criteria. He then added data from the top 170 competitors at the 2001–2002 British Indoor Rowing Championships. The mean ergometer profile was 101.5%, 99.8%, 99.0%, 99.7% of mean race speed across the four 500 m sectors — a fast first 500 m, a slowest third 500 m, and a small finishing lift. The on-water profile was more front-loaded: 103.3%, 99.0%, 98.3%, 99.7%. The pacing profile did not discriminate between winners and losers, or between men and women, so the fast first 500 m is not by itself a marker of a superior race ([1] Garland 2005, Level 4; [15] Hagerman 1984, Level 5).

The read is direct: an even split is not the empirical pattern of elite 2K rowers. The pattern is a controlled-positive first 500 m, a back half at or just below average, and a small finishing lift. A strict negative split — second half faster than the first — is not the empirically observed strategy of elite 2K rowers, and there is no peer-reviewed rowing-specific evidence that it outperforms a reverse-J pacing profile in trained rowers ([1] Garland 2005, Level 4). For 5K and 6K the same broad pattern holds but the back half is closer to the average and the finishing lift is smaller — the longer the distance, the less front-loading the plan ([7] Astridge et al. 2024, Level 2b; [11] World Rowing — Indoor rules, Level 5).

The habituation-trial problem: trial 1 is a pacing-learning trial

The 2013 [2] Gee, French, Gibbon, and Thompson paper is the second foundational anchor (Level 2b). Fourteen well-trained male rowers did three 2K trials on a Concept2 ergometer at three-to-seven-day intervals, with power, stroke rate, heart rate, and blood lactate measured per 500 m. Performance time was unchanged across trials (typical error 2.4%), but pacing strategy shifted substantially: trial 1 was a positive split (higher starting power, progressive decrease), while trials 2 and 3 settled into a reverse-J — conservative start, end-spurt with elevated power in the final 500 m. The authors concluded that intervention studies using 2K performance need a habituation trial before reliable measurement, and the rower who has only done one or two 2Ks is leaving several seconds on the piece through pacing alone ([2] Gee et al. 2013, Level 2b).

The implication for the AI coach is direct. A 2K rowed for the first time after a long layoff is partially a pacing-learning trial, not a calibration of physiology. The rower who has done five or more 2Ks has a stable pacing pattern and is testing physiology. The coach that anchors reference pace on the moving average of the most recent two-to-three 2Ks — and treats the first 2K after a layoff as a familiarisation piece — is reading pacing correctly. The coach that anchors on a single recent 2K without considering whether the rower has done it recently is reading pacing incorrectly ([2] Gee et al. 2013, Level 2b; [5] Schabort et al. 1999, Level 2b).

The degressive strategy is not the optimal strategy for trained rowers

The 2022 [3] Boillet, Haas, Samozino, Morel, Bowen, Cohen, and Messonnier paper is the most direct experimental test of pacing strategies in trained rowers (Level 2b). Thirteen French rowers of national and ex-international level (4 women, 9 men) performed a simulated 2K on a Concept2 ergometer with simultaneous measurement of handle force/velocity, VO2, HR, blood lactate, SpO2, RPE, and computed power, energy, and efficiency, under three pacing strategies for the first 1500 m: degressive (fast-start, progressive decline), progressive (slow-start, accelerating), and stable (constant).

The finding was clear: the degressive strategy was the least-preferred strategy in the post-trial questionnaire, and produced more time at high blood lactate and high RPE than the stable or progressive strategies. Final blood lactate concentration and final RPE at the end of the first 1500 m did not differ significantly across strategies, but lactate and RPE rose sooner under the degressive strategy, so the rowers spent more of the trial in the high-strain zone ([3] Boillet et al. 2022, Level 2b). Oxidative energy contribution did not differ significantly across strategies. The mechanistic read: an aggressive opening is not free, even when the final physiological values look similar — the rower pays for it earlier in the piece.

The implication for the rower is direct. The peer-reviewed experimental evidence does not support the common coaching claim that "go out hard to set up the back half." The degressive strategy is empirically common in 2K rowers, but the 2022 Boillet data shows it is the least-preferred strategy in trained rowers, and the strain profile it produces is worse, not better ([3] Boillet et al. 2022, Level 2b). For the AI coach that recommends a pace plan, the read is: do not default to a hard first 500 m. A controlled-positive first 500 m — a few seconds faster than average, not a sprint — is the empirically observed pattern of elite 2K rowers, and it produces less early strain than a degressive opening ([1] Garland 2005, Level 4; [3] Boillet et al. 2022, Level 2b; [21] Riechman et al. 2002, Level 2b; [22] Cerasola et al. 2020, Level 2b).

The fast-start mechanism: priming, VO2 kinetics, and the boundary of the evidence

The mechanistic case for a fast start rests on a different body of evidence. The 2014 [8] Turnes, Salvador, Lisbôa, de Aguiar, Cruz, and Caputo paper (Level 2b) put eleven active men (VO2max 57 ± 4 mL/kg/min) through treadmill-running trials at the boundary of the severe-intensity domain. A fast start reduced mean VO2 response time to 19.3 s versus 22.2 s for constant pace (P = 0.025), and improved two supramaximal-performance measures (matched-duration distance and matched-speed tolerance). The 2018 [4] Brock, Antonellis, Black, DiMenna, Vanhatalo, Jones, and Bailey paper (Level 2b) extended this in cycling: nine men did a 4-km cycling time trial (~114 ± 17 kJ) under four conditions (all-out primed, all-out unprimed, self-paced primed, self-paced unprimed). The all-out-primed condition cut VO2 mean response time to 20 ± 6 s versus 42 ± 13 s for self-paced unprimed, and shortened trial-completion time to 402 ± 14 s versus 411 ± 16 s (P < .05).

The mechanism is real and transfers conceptually to rowing: an aggressive opening accelerates pulmonary VO2 adjustment, and the speed of VO2 adjustment is a real lever on severe-intensity performance ([8] Turnes et al. 2014, Level 2b; [4] Brock et al. 2018, Level 2b). The boundary of the evidence is also real: both studies were in running or cycling, not rowing; both populations were small (11 and 9 men); both authors explicitly bounded the conclusion to efforts of 2–3 minutes duration, well below the 6–8 minute 2K and far below 5K/6K; and the Brock 2018 result required a priming exercise preceding the all-out start — without priming, the all-out start alone did not produce a statistically significant benefit over self-paced ([4] Brock et al. 2018, Level 2b).

The implication for the rower is honest. A controlled-positive first 500 m of a 2K probably does recruit the VO2-kinetics advantage demonstrated in the cycling and running literature — and the manufacturer's warm-up guidance is consistent with the priming requirement ([13] Concept2 — Warm-up blog, Level 5). At the same time, the 2K literature has its own counter-evidence: a well-controlled acute intervention (ischaemic preconditioning) failed to shift 2K time in trained rowers ([23] Turnes et al. 2018, Level 2b), so a controlled-positive first 500 m should not be assumed to deliver a measurable performance gain — it is a pacing plan, not a magic lever. A degressive opening that is substantially faster than the empirical elite mean of 101.5% of average race speed is paying for the VO2-kinetics benefit with earlier physiological strain ([3] Boillet et al. 2022, Level 2b). The optimal trade-off for a trained rower is a controlled-positive first 500 m (a few seconds faster than average, not a sprint) preceded by a warm-up that primes the aerobic system ([9] Concept2 — Pacing blog, Level 5; [10] Concept2 — 2K test blog, Level 5; [13] Concept2 — Warm-up blog, Level 5).

Pacing variability and the role of experience

Pacing variability — the split-to-split oscillation in stroke rate and power output across the piece — differs by experience. The most experienced rowers show the smallest within-trial variability: a familiarised rower holds a tighter band around the target split, while a first-time tester oscillates more, particularly in the back half when fatigue accumulates ([2] Gee et al. 2013, Level 2b). The Garland 2005 elite-2K data is consistent: the mean 500 m sectors of the top 170 British Indoor Rowing Championship competitors cluster within a few percent of the average, and the standard deviation around the mean 500 m sector profile is small ([1] Garland 2005, Level 4).

The implication for the AI coach is direct. A rower's pacing variability is itself a signal: a rower whose 2K shows large split-to-split oscillation is either under-familiarised with the test or under-trained on sustained power output. The coach that reads the variability alongside the average split is reading the rower more honestly than the coach that reads the average alone. The coach that updates reference pace on a moving average of the most recent two-to-three 2Ks is implicitly averaging over both the split oscillation and the mean shift, which is the right thing to do for familiarised rowers ([2] Gee et al. 2013, Level 2b; [5] Schabort et al. 1999, Level 2b; [6] Ingham et al. 2002, Level 2b).

For the rower, the practical read is: do the same 2K three times before treating the result as a calibration. The first trial is a pacing-learning trial; the second and third are calibration trials; the moving average of the second and third is the most reliable signal. A rower who has done two 2Ks a few days apart and sees a five-second-per-500 m difference is inside the noise band of a familiarisation shift and should not over-interpret it ([2] Gee et al. 2013, Level 2b; [5] Schabort et al. 1999, Level 2b).

Negative splits: when they work and when they fail

A negative split — second half faster than the first half — is the strategy most often recommended in older coaching literature, especially for longer pieces. The peer-reviewed rowing evidence does not show that elite 2K competitors use a strict negative split ([1] Garland 2005, Level 4); the empirical elite 2K profile is a controlled-positive first 500 m with a small finishing lift, not a strictly faster second half. The mechanistic case for negative splitting rests on the VO2 slow component — the additional oxygen cost of exercise at a constant workload above the lactate threshold that develops over minutes — and on the fatigue-resistance argument that a slower first half preserves glycogen and reduces lactate accumulation for the back half.

The case is plausible but does not transfer cleanly to the 2K. The 2K is short enough that the VO2 slow component has only minutes to develop, and short enough that the practical back-half benefit of a slower first half is small. For 5K and 6K the case is stronger: a slower first half preserves more for the back half and the slow-component cost is larger, so a controlled-positive or near-even pacing profile with a finishing lift fits the literature better than a strict negative split ([7] Astridge et al. 2024, Level 2b; [1] Garland 2005, Level 4; [11] World Rowing — Indoor rules, Level 5). For a 2K rowed by a novice, a strict negative split is rarely the right plan: the novice has not yet learned what the back-half split should be, and the first 500 m is the only one the rower can control with confidence.

The practical read: a strict negative split is a strategy for rowers who know their target split well enough to row the first half conservatively and trust the back half. For a novice, the right strategy is even or slightly positive. For an intermediate, the right strategy is controlled-positive with a finishing lift. For an advanced rower who knows the target split well, the right strategy is a controlled-positive first 500 m, a middle at or just below average, and a deliberate end-spurt — and a strict negative split is rarely the empirical choice even at the elite level ([1] Garland 2005, Level 4; [2] Gee et al. 2013, Level 2b; [3] Boillet et al. 2022, Level 2b; [9] Concept2 — Pacing blog, Level 5; [10] Concept2 — 2K test blog, Level 5).

Practical recommendations by rower level

The peer-reviewed literature and the manufacturer-and-federation guidance converge on a tiered practical recipe.

Novice (first 2–3 2Ks ever, or first 2K after a long layoff). The right plan is even or slightly positive. The first 2K is a pacing-learning trial; the rower is learning what the target split feels like, and a strictly negative split or a strict end-spurt requires knowing the target split in advance. The 2022 Boillet data shows that aggressive opening strategies produce earlier physiological strain in trained rowers, and the effect is likely larger in novices who have not yet learned to tolerate the strain ([3] Boillet et al. 2022, Level 2b; [9] Concept2 — Pacing blog, Level 5; [10] Concept2 — 2K test blog, Level 5; [12] British Rowing — Go Row Indoor tests, Level 5).

Intermediate (4–10 2Ks done, comfortable with the target split). The right plan is controlled-positive with a deliberate end-spurt. The first 500 m is a few seconds faster than the average, the middle is at or just below the average, and the final 500 m is a few seconds faster again. This is the empirical pattern of elite 2K rowers ([1] Garland 2005, Level 4) and the strategy the manufacturer and federation guidance converges on ([9] Concept2 — Pacing blog, Level 5; [12] British Rowing — Go Row Indoor tests, Level 5). The intermediate rower has done enough 2Ks to trust the target split and to plan a controlled-positive opening that does not pay for the VO2-kinetics benefit with excessive early strain.

Advanced (10+ 2Ks done, knows the target split well, trains consistently). The right plan is a controlled-positive reverse-J — a fast first 500 m (a few seconds faster than the target, not a sprint), the middle at or just below the target, and a deliberate end-spurt in the final 500 m that may be the fastest 500 m of the piece. The 2013 Gee data shows that well-trained rowers converge on this shape after familiarisation ([2] Gee et al. 2013, Level 2b), and the 2005 Garland data shows this is the elite 2K profile ([1] Garland 2005, Level 4). The advanced rower has done enough 2Ks to control the front-loading — getting the first 500 m fast enough to recruit the VO2-kinetics advantage without paying for it with excessive early strain — and to deliver the end-spurt without catastrophic fade in the back half. The advanced rower should also treat controllable test-day variables — sleep, hydration, damper — as part of the pace plan, since dehydration of even 2% body-mass can shorten time-to-exhaustion substantially ([25] Ebert et al. 2007, Level 2b; [14] Concept2 — Damper setting blog, Level 5).

The 5K and 6K: how pacing changes with distance

For longer pieces the broad pattern holds but the back half moves closer to the average and the finishing lift is smaller. The 2024 [7] Astridge et al. paper (Level 2b) compared 2000 m and 1500 m pacing on the ergometer and showed the pacing profile differs in shape and in the relative weight of the end-spurt. The 5K and 6K data points are sparser in the peer-reviewed literature specifically for the indoor ergometer, but the broader cycling and running time-trial literature converges on a similar pattern: the longer the distance, the less front-loaded the optimal pace plan, and the larger the slow-component cost of a fast start ([8] Turnes et al. 2014, Level 2b; [4] Brock et al. 2018, Level 2b).

The practical read for the indoor rower: for a 5K or 6K on the Concept2 ergometer, a controlled-positive first 500 m with a near-even middle and a smaller finishing lift is the right shape. A strict negative split is rare at the elite level even for shorter pieces, and is even rarer for longer pieces — the rower who tries to row a strict negative split on a 5K risks a back half that is physiologically impossible at the split the first half suggested ([1] Garland 2005, Level 4; [7] Astridge et al. 2024, Level 2b; [11] World Rowing — Indoor rules, Level 5).

What the AI coach actually does with pacing

For an AI coach that reads the rower's logbook and writes the rower's session, pacing is one of the controllable variables the coach can shape directly. The coach's rule is:

• A rower who has done fewer than three 2Ks — the coach widens the reference-pace band and leans on rate caps, recent training data, and the rower's stated context. The first 2K is a pacing-learning trial, not a calibration of physiology ([2] Gee et al. 2013, Level 2b).
• A rower who has done three or more 2Ks but the most recent is more than six weeks old — the coach treats the most recent 2K as a familiarisation piece and waits for a fresh test before updating the reference pace. The pacing pattern may have shifted during the layoff ([2] Gee et al. 2013, Level 2b).
• A rower who has done three or more 2Ks with the most recent within six weeks — the coach trusts the moving average of the most recent two-to-three 2Ks and uses it as the reference pace for the next session. This is the calibration signal ([5] Schabort et al. 1999, Level 2b; [2] Gee et al. 2013, Level 2b).
• A rower whose most recent 2K shows large split-to-split oscillation — the coach notes the variability and prescribes a familiarisation session (an easy 4×500 m at target split with full rest) before the next calibration 2K. The variability is itself a signal ([2] Gee et al. 2013, Level 2b).
• A rower who wants to use a strict negative split — the coach notes the strategy but does not default to it; the empirical pattern of elite 2K rowers is controlled-positive or reverse-J, not strictly negative ([1] Garland 2005, Level 4; [3] Boillet et al. 2022, Level 2b).
• A rower with no recent 2K but a 5K in the logbook — the coach estimates a 2K-equivalent split from the 5K using a power-law cross-distance prediction (the canonical Riegel-style formula T₂ = T₁ × (D₂/D₁)^1.06 is the standard starting point, but Concept2 publishes its own cross-distance regression tables), with a wider confidence band than a directly-rowed 2K. The reference pace widens, but the rower still has a signal. Concept2 publishes the cross-distance prediction as part of its ranking methodology ([5] Schabort et al. 1999, Level 2b; [6] Ingham et al. 2002, Level 2b; [27] Concept2 — Rankings, Level 5; [28] Concept2 — Ranking help, Level 5).

The four-numbers dashboard (CTL, ATL, TSB, ACWR) is the how much the coach is asking the body to absorb in a given week. The 2K trajectory is the where the coach expects the body to be in a few months. The reference pace is the today. The pacing strategy is the in-test plan that connects the today to the where, and the rower's level of familiarisation with the test pace is the signal the coach reads to choose the right plan ([2] Gee et al. 2013, Level 2b; [16] Seiler & Kjerland 2006, Level 4; [17] Rosenblat et al. 2025, Level 2a; [18] Sylta et al. 2014, Level 4; [19] Bourdon et al. 2009, Level 2b; [24] Otter et al. 2015, Level 2b).

Limitations and open questions

The rowing-specific pacing literature is small. Most of the foundational papers are in cycling, running, or kayaking, with rowing-specific work concentrated on the 2K and the 14-rower Gee et al. sample ([2] Gee et al. 2013, Level 2b). The 13-rower Boillet et al. sample is the most direct experimental test of pacing strategies in trained rowers ([3] Boillet et al. 2022, Level 2b). The 5K and 6K rowing-specific pacing literature is sparser still. The reader should weight the rowing-specific evidence more heavily than the cross-sport evidence when the two diverge.

The degressive-vs-controlled-positive comparison is bounded. The 2022 [3] Boillet paper compared degressive, progressive, and stable strategies for the first 1500 m of a 2K — not the full 2000 m. The end-spurt was held constant across conditions. The reader who wants to know whether a controlled-positive first 500 m followed by a reverse-J back half outperforms a strictly faster first 500 m followed by a slower back half will not find that experiment in the published literature. The Garland 2005 elite-2K profile is observational race data, not an intervention trial ([1] Garland 2005, Level 4).

The fast-start mechanism evidence is from cycling and running. Both the 2014 Turnes paper and the 2018 Brock paper are in non-rowing modalities, both in small male samples (11 and 9 men), and both explicitly bounded to efforts of 2–3 minutes duration ([8] Turnes et al. 2014, Level 2b; [4] Brock et al. 2018, Level 2b). The mechanism (faster VO2 adjustment) transfers conceptually to rowing, but the effect size in a 6–8 minute 2K is unknown.

The sex-specific and age-specific responses are not as well characterised as the male response. The 2003 [26] Yoshiga and Higuchi paper (Level 2b) provided sex-specific 2K-physiology data, but the underlying literature on female rowing pacing is much smaller than the male literature ([26] Yoshiga & Higuchi 2003, Level 2b). The within-sex variance is large enough that a single 2K does not pin down the rower's optimal pacing strategy, and the AI coach that uses a single test's pacing profile as the rower's "true" profile is over-fitting.

The pacing recommendations are bounded to indoor rowing. The on-water 2K profile is more front-loaded than the ergometer profile (Garland 2005: 103.3% vs 101.5% on the first 500 m), so a strategy that works on the Concept2 ergometer does not transfer directly to on-water pacing ([1] Garland 2005, Level 4). The reader who rows both should treat the ergometer and on-water pace plans as distinct.

What to do with this article

Read the principle: pacing is the controllable variable on top of physiology in a 2K or 5K, and the peer-reviewed literature shows that the empirical pattern of elite 2K rowers is a controlled-positive first 500 m with a small finishing lift — not a strictly even split and not a strict negative split. Read the evidence: the [1] Garland 2005 (Level 4) elite-2K profile paper shows the empirical shape; the [2] Gee et al. 2013 (Level 2b) pacing-consistency paper shows trial 1 is a familiarisation piece; the [3] Boillet et al. 2022 (Level 2b) pacing-strategy paper shows the degressive strategy produces earlier strain in trained rowers; the [4] Brock et al. 2018 (Level 2b) priming + all-out-start paper shows the VO2-kinetics mechanism transfers to severe-intensity exercise. Read the practical read: novice = even or slightly positive; intermediate = controlled-positive with end-spurt; advanced = reverse-J with deliberate end-spurt; 5K/6K = less front-loaded, smaller finishing lift.

When you want to write a pace plan for your next 2K, the practical recipe is: row a familiarisation piece first if it has been more than six weeks since your last 2K, then row your calibration piece with the pace plan that fits your level; treat the first 2K after a layoff as a pacing-learning trial, not a calibration of physiology; trust the moving average of your most recent two-to-three 2Ks more than any single test; and let the AI coach read the pacing pattern alongside the average split, because the variability is itself a signal.