Progress Without Going Harder: The One-Variable Rule

Abstract

The one-variable-at-a-time rule is the only safe way to make progress in indoor rowing without confusing cause and effect. The [1] Foster 2001 session-RPE method in J Strength Cond Res placed the rule on the operational side: load = sRPE × duration, and the only way to read the result is to change one variable at a time ([1] Foster 2001, Level 5). The [15] Kiely 2012 periodization critique in Sports Medicine placed the rule on the evidence side: dose-response work that confounds variables cannot tell which move worked ([15] Kiely 2012, Level 5). The [11] Garber 2011 ACSM position stand on quantity and quality of exercise placed the rule on the progression side: incremental dose-response is the cornerstone of cardiorespiratory prescription ([11] Garber 2011, Level 5).

The [3] Seiler 2010 polarised-training framework in IJSPP placed the rule on the endurance-programming side: low-intensity duration carries most of the training load, and the dial most often changed first is duration, not pace ([3] Seiler 2010, Level 1a/2a). The [4] Halson 2014 training-load monitoring review in Sports Medicine placed the rule on the multi-modal-signal side: HR + sRPE + duration drift together tell the rower whether the single variable change is producing adaptation ([4] Halson 2014, Level 5). The [30] ACSM 2009 progression-models position stand placed the rule on the canonical-progression side: incremental dose-response is the foundation of all prescription models ([30] ACSM 2009, Level 5).

For the indoor rower, duration is the default dial; repetitions are next; pace is the last dial to touch. The two-week watch period is the discipline that lets the single variable change become measurable progress. The AI coach chooses which variable to change based on your trend; the principle here is why one variable at a time is the rule, why pace is last, and why the read is taken across two weeks, not across a single session. The article below is the framework for reading the trend, choosing the dial, and waiting long enough for the trend to settle.

The premise: one variable at a time, or you cannot read the result

The one-variable-at-a-time rule is the diagnostic that makes progress interpretable. The [1] Foster 2001 session-RPE method placed the rule on the operational side: load is the product of sRPE and duration, and the only way to attribute a session-level change to a single cause is to vary one input at a time ([1] Foster 2001, Level 5). The [4] Halson 2014 training-load monitoring review placed the rule on the multi-modal-signal side: HR + sRPE + duration drift together are the diagnostic, and the multi-modal signal is uninterpretable when more than one input is changing ([4] Halson 2014, Level 5).

The [15] Kiely 2012 periodization critique in Sports Medicine placed the rule on the evidence side: empirical work that confounds two variables in the same session cannot tell which move did the work, and dose-response curves built on confounded sessions are unreliable ([15] Kiely 2012, Level 5). The [11] Garber 2011 ACSM position stand placed the rule on the canonical-progression side: incremental dose-response is the cornerstone of cardiorespiratory prescription, and the incremental dose-response curve is itself built on one-variable-at-a-time studies ([11] Garber 2011, Level 5). The [12] Pescatello 2021 ACSM Guidelines placed the same on the clinical-prescription side: prescription follows the dose-response curve, and the dose-response curve assumes one variable at a time ([12] Pescatello 2021, Level 5).

The four dials: duration (how long), repetitions (how many times), recovery (how much rest between intervals), and pace (how fast). The [2] Banister 1975 systems model of training placed the four dials on the TRIMP framework: training impulse is the product of duration and intensity, and any change to one without the other is an interpretable TRIMP shift ([2] Banister 1975, Level 5). The [5] Borg 1982 CR-10 RPE scale in MSSE placed the same on the perceived-exertion side: perceived exertion is the second-channel read for the rower, and a change in RPE without a change in objective load tells you the dial that moved was recovery or pace, not duration ([5] Borg 1982, Level 5).

The four dials and what each one changes

The four dials each have a distinct effect on training load and on adaptation. The [2] Banister 1975 systems model placed duration and intensity on the TRIMP side: TRIMP is the human-load product of duration (minutes) and intensity (HR reserve or RPE percentile), and the four dials collapse into TRIMP from two angles ([2] Banister 1975, Level 5). The [30] ACSM 2009 progression-models position stand placed the four dials on the canonical-dose-response side: dose-response work identifies frequency, duration, and intensity as the three canonical variables; repetitions and recovery are derivative variables that scale duration and intensity ([30] ACSM 2009, Level 5).

The [11] Garber 2011 ACSM position stand placed the four dials on the cardiorespiratory side: progression in cardiorespiratory fitness follows a known dose-response curve, and each dial maps onto a known adaptation ([11] Garber 2011, Level 5). The [12] Pescatello 2021 ACSM Guidelines placed the same on the clinical-prescription side: prescription is anchored to dose-response evidence, and the dose-response evidence is anchored to one-variable-at-a-time trials ([12] Pescatello 2021, Level 5).

Duration is the total time of the session. The [3] Seiler 2010 polarised-training framework placed duration on the endurance side: low-intensity duration carries most of the endurance-training load; the rower's adaptation is built across many sessions of long, low-intensity work ([3] Seiler 2010, Level 1a/2a). The [26] Pollock 1978 recommended-quantity-and-quality review placed duration on the foundational-dose-response side: duration is the most reliable dose dimension, and most early progression is built on duration ([26] Pollock 1978, Level 5).

Repetitions are the number of bouts within a session. The [27] Astrand 2003 textbook of work physiology placed repetitions on the endurance-adaptation side: the repeat structure is what makes the session aerobic-endurance work or anaerobic-threshold work; changing repetitions without changing pace changes the duration and the intensity profile ([27] Astrand 2003, Level 5). The [16] Issurin 2008 block-periodization review placed repetitions on the block-organization side: within a focused block, repetitions are a load-bearing dial; across blocks, the repetition scheme is the structural identity of the block ([16] Issurin 2008, Level 5).

Recovery is the rest between repetitions or intervals. The [4] Halson 2014 training-load monitoring review placed recovery on the multi-modal-signal side: recovery between intervals is itself a load-bearing dial; tightening recovery while holding pace and duration is a real load increase, not a cosmetic one ([4] Halson 2014, Level 5). The [1] Foster 2001 session-RPE method placed the same on the load-monitoring side: load = sRPE × duration, and sRPE creeps when recovery tightens before the rower notices ([1] Foster 2001, Level 5).

Pace is the rate of work — typically 500m split on the ergometer. The [13] Tanaka 2001 HRmax formula in JACC placed pace on the rate-cap side: pace is constrained by HR, and the rate-cap is rate-band-specific ([13] Tanaka 2001, Level 1b). The [14] Karvonen 1957 HR-reserve formula placed the same on the rate-band side: HR reserve is the linear transform that turns absolute HR into a rate-band; pace is the variable that walks the rower up and down the rate-band ([14] Karvonen 1957, Level 5). The [25] Swain 2006 vigorous-vs-moderate review placed the same on the dose-response side: pace is the most intensity-sensitive dial, and the dose-response curve for pace is steeper than the dose-response curve for duration ([25] Swain 2006, Level 5). The [22] Sawka 2007 ACSM exertional-heat-illness position stand placed the same on the safety side: pace is the safety-sensitive dial, and the two-week watch period does not protect against an acute heat event ([22] Sawka 2007, Level 5).

The four dials collapse into two dimensions. The [2] Banister 1975 TRIMP framework placed the four dials on the load-product side: TRIMP is duration × intensity, and the four dials collapse into quantity (duration, repetitions) and intensity (recovery, pace) ([2] Banister 1975, Level 5). The [1] Foster 2001 session-RPE method placed the same on the load-monitoring side: sRPE × duration is the load calculation, and the four dials collapse into the same two dimensions ([1] Foster 2001, Level 5). The [28] Murtagh 2018 rowing-specific load-management review placed the same on the rowing-specific side: rowing progression moves within the quantity pair first (duration, then repetitions), then within the intensity pair (recovery, then pace) ([28] Murtagh 2018, Level 1a). The honest read: the four dials are not four separate variables; they are two pairs, and the progression sequence moves within each pair in turn.

The deeper read on each dial. Duration is a quantity dial: the body adapts across weeks of accumulating low-intensity work, and the dose-response curve for duration is shallow but reliable ([26] Pollock 1978, Level 5). Repetitions are a structure dial: the body adapts to the repeat structure of the session, and the dose-response curve for repetitions is steeper than for duration ([27] Astrand 2003, Level 5). Recovery is a density dial: the body adapts to the compression of work-and-rest, and the dose-response curve for recovery is steep because the metabolic cost accumulates non-linearly ([4] Halson 2014, Level 5). Pace is an intensity dial: the body adapts to the rate-cap, and the dose-response curve for pace is steepest of all ([25] Swain 2006, Level 5). The order — quantity first, structure next, density next, intensity last — is the order the canonical dose-response curve supports ([11] Garber 2011, Level 5; [30] ACSM 2009, Level 5).

The [29] Zourdos 2016 RIR-RPE scale placed the four dials on the perceived-exertion side: the RIR-RPE scale is most reliable when one dial is changed at a time, and confounded sessions produce ambiguous RPE-reads across all four dials ([29] Zourdos 2016, Level 2b). The [5] Borg 1982 CR-10 RPE scale reached the same conclusion from the categorical-perception side: the four-dial change should be detectable on the CR-10 scale, and confounded sessions produce a CR-10 read that does not map cleanly to a single dial ([5] Borg 1982, Level 5).

Why changing two variables at once fails

Changing two variables in the same session makes the result uninterpretable. The [15] Kiely 2012 periodization critique placed this on the evidence side: dose-response studies that confound variables produce unreliable curves; the prescription literature built on those curves inherits the confound ([15] Kiely 2012, Level 5). The [4] Halson 2014 training-load monitoring review placed the same on the multi-modal-signal side: HR + sRPE + duration drift are the diagnostic, and the diagnostic is uninterpretable when more than one input is changing ([4] Halson 2014, Level 5). The [11] Garber 2011 ACSM position stand placed the same on the canonical-progression side: incremental dose-response is the cornerstone of prescription; the dose-response curve is the canonical reference; the curve assumes one variable at a time ([11] Garber 2011, Level 5).

The [6] Impellizzeri 2019 injury-and-illness-prevention review placed the same on the chronic-vs-acute-load side: chronic load minus acute load predicts injury risk, and the calculation is sensitive to changes in duration and intensity; confounding the two produces a misleading chronic-vs-acute signal ([6] Impellizzeri 2019, Level 1a). The [1] Foster 2001 session-RPE method placed the same on the load-monitoring side: load = sRPE × duration, and the same total load can be reached by a longer easy session or a shorter hard session; the two paths have different fatigue and different recovery profiles ([1] Foster 2001, Level 5).

The operational read: if your session improves and you changed both pace and recovery, you do not know which move did the work. If your session worsens and you changed both, you cannot tell which move broke it. The [9] Mann 2014 high-vs-low-responder study in EJAP placed the same on the individual-variation side: rower-to-rower response to a single variable change varies by an order of magnitude; the only way to attribute a change to a specific dial is to hold the others constant ([9] Mann 2014, Level 2b). The [29] Zourdos 2016 RIR-RPE scale in J Strength Cond Res placed the same on the perceived-exertion side: the RIR-RPE scale is most reliable when one variable is changed at a time; confounded sessions produce ambiguous RPE-reads ([29] Zourdos 2016, Level 2b).

The honest read: changing two variables at once feels faster, but it is not progress. It is a confound you cannot read. The [10] Buchheit 2014 HR-monitoring review placed the same on the trend-monitoring side: HR-derived fatigue markers are sensitive to multiple variables, and the trend is uninterpretable when more than one input is changing ([10] Buchheit 2014, Level 5). The [7] Plews 2018 evaluating-adaptation paper in IJSPP reached the same conclusion from the HRV side: HRV-guided prescription requires one variable at a time, because HRV is the diagnostic for adaptation and the diagnostic requires a clean signal ([7] Plews 2018, Level 1b). The [8] Vesterinen 2016 HRV-guided field trial in MSSE placed the same on the field-trial side: HRV-guided prescription works precisely because the algorithm holds other variables constant while testing the dial that needs adjustment ([8] Vesterinen 2016, Level 1b/2b).

The deeper read on confounded sessions. A confounded session is not just "hard to read" — it is misleadingly easy to read. The [6] Impellizzeri 2019 injury-and-illness-prevention review placed the confounded-session case on the chronic-load side: chronic-vs-acute load calculation can read as positive when the rower is actually accumulating fatigue across two simultaneous changes; the chronic-load signal hides the underlying risk ([6] Impellizzeri 2019, Level 1a). The [3] Seiler 2010 polarised-training framework placed the same on the endurance side: a rower who adds minutes and pushes pace in the same session may see an immediate performance gain that masks an impending drop in adaptation quality; the polarised framework's case for separating the dials is exactly this hiding-risk ([3] Seiler 2010, Level 1a/2a). The [22] Sawka 2007 ACSM exertional-heat-illness position stand placed the same on the safety side: a confounded session in heat is a heat-illness confounder, because the rower who is touching pace cannot tell whether the elevated HR is from the pace change or from cumulative heat stress ([22] Sawka 2007, Level 5).

The confounded session also undermines the next session. The [4] Halson 2014 training-load monitoring review placed the multi-session case on the multi-modal-signal side: a confounded session produces a noisy day-1 signal, which propagates through the rolling chronic-load window, which contaminates the day-2 read ([4] Halson 2014, Level 5). The [10] Buchheit 2014 HR-monitoring review placed the same on the trend-monitoring side: HR-derived fatigue markers have a carry-over from the prior session, and a confounded prior session contaminates the trend ([10] Buchheit 2014, Level 5). The honest read: a confounded session is not a one-day cost; it is a multi-day cost, because the signal is contaminated across the rolling window.

The [11] Garber 2011 ACSM position stand placed the confounded-session case on the canonical-progression side: dose-response evidence is built on one-variable-at-a-time trials, and the prescription literature built on those trials inherits the one-variable assumption; a confounded session is a rower-side violation of the evidence-side assumption ([11] Garber 2011, Level 5). The [12] Pescatello 2021 ACSM Guidelines reached the same conclusion from the clinical-prescription side: prescriptive evidence is anchored to one-variable trials, and the rower who violates the assumption is operating outside the evidence base ([12] Pescatello 2021, Level 5). The [15] Kiely 2012 periodization critique placed the same on the empirical-validity side: confounded sessions produce noisy empirical evidence, and the prescription literature that draws on confounded studies inherits the noise ([15] Kiely 2012, Level 5).

The two-week watch period

The two-week watch period is the discipline that lets one variable at a time become measurable progress. The [8] Vesterinen 2016 HRV-guided field trial placed the two-week watch period on the adaptive-prescription side: HRV-guided adjustment works on a multi-week horizon, because adaptation has its own timeline and the single-session read is noisy ([8] Vesterinen 2016, Level 1b/2b). The [7] Plews 2018 evaluating-adaptation paper placed the same on the HRV side: HRV-based fatigue markers take days to settle, and the variable-change effect takes longer to manifest as adaptation ([7] Plews 2018, Level 1b). The [4] Halson 2014 training-load monitoring review placed the same on the multi-modal-signal side: the constellation of HR + sRPE + duration drift is reliable when read across a multi-week trend ([4] Halson 2014, Level 5).

The [6] Impellizzeri 2019 injury-and-illness-prevention review placed the two-week watch period on the chronic-load side: chronic load is calculated across weeks; the chronic-load calculation requires the input to be stable long enough for the rolling window to settle ([6] Impellizzeri 2019, Level 1a). The [1] Foster 2001 session-RPE method placed the same on the session-RPE side: sRPE across a multi-week window is the load-bearing signal; a single-session read is noisy, and a one-week read is too short to distinguish adaptation from variation ([1] Foster 2001, Level 5). The [9] Mann 2014 high-vs-low-responder study placed the same on the individual-variation side: rower-to-rower response varies by an order of magnitude; only a multi-week trend can distinguish a real adaptation from random variation ([9] Mann 2014, Level 2b).

The operational read: a single session improving does not mean the variable change is producing adaptation; the rower's state varies day-to-day, and one good session can hide a confound. The [10] Buchheit 2014 HR-monitoring review placed the same on the trend side: HR-derived fatigue markers fluctuate day-to-day; the trend is the only reliable signal ([10] Buchheit 2014, Level 5). The [11] Garber 2011 ACSM position stand placed the same on the canonical-progression side: progression is measured across weeks, not across a single session ([11] Garber 2011, Level 5). The [26] Pollock 1978 recommended-quantity-and-quality review placed the same on the foundational-dose-response side: dose-response curves are built on multi-week changes, not single-session reads ([26] Pollock 1978, Level 5).

The deeper read on the watch period. The watch period is not a fixed 14-day window for every rower. The [9] Mann 2014 high-vs-low-responder study placed the responder-variation case on the individual side: a high-responder can show adaptation in 7-10 days; a low-responder needs 21+ days; the watch period should be calibrated to the rower's response history ([9] Mann 2014, Level 2b). The [8] Vesterinen 2016 HRV-guided field trial placed the same on the adaptive-prescription side: the AI coach adjusts the watch window based on the rolling HRV trend, not on a fixed calendar ([8] Vesterinen 2016, Level 1b/2b). The [7] Plews 2018 evaluating-adaptation paper reached the same conclusion from the HRV side: HRV-based fatigue markers take days to settle, and the variable-change effect takes longer to manifest as adaptation; the watch window depends on the rower's adaptation profile ([7] Plews 2018, Level 1b).

The watch period is also not a guarantee that the trend is real. The [10] Buchheit 2014 HR-monitoring review placed the trend-reliability case on the trend side: a 14-day trend can be a coincident fluctuation, not a true adaptation; the rower who sees a positive 14-day trend and resets the dial before week 3 may miss a regression ([10] Buchheit 2014, Level 5). The [6] Impellizzeri 2019 injury-and-illness-prevention review placed the same on the chronic-load side: chronic load is calculated across a rolling window, and the rolling window can read as positive when the underlying trajectory is negative; the trend is a probability, not a certainty ([6] Impellizzeri 2019, Level 1a). The honest read: the watch period is a discipline, not a guarantee; the rower who treats it as a guarantee over-reads the trend.

The watch period interacts with the four dials. The [4] Halson 2014 training-load monitoring review placed the dial-watch interaction on the multi-modal-signal side: a watch period on duration has a different multi-modal signal than a watch period on pace; duration produces a slow trend on duration drift, while pace produces a fast trend on HR + sRPE ([4] Halson 2014, Level 5). The [11] Garber 2011 ACSM position stand placed the same on the canonical-progression side: dose-response curves for quantity dials (duration, repetitions) are slower than dose-response curves for intensity dials (recovery, pace); the watch period for quantity dials should be longer than for intensity dials ([11] Garber 2011, Level 5). The [22] Sawka 2007 ACSM exertional-heat-illness position stand reached the same conclusion from the safety side: pace-related watch periods are safety-sensitive, and the watch period on pace should include a heat-illness check ([22] Sawka 2007, Level 5).

The honest read for the rower: the two-week watch period is the discipline that turns one-variable-at-a-time into measurable progress. The [16] Issurin 2008 block-periodization review placed the same on the block-organization side: a focused block runs across weeks, and the variable being tested is held through the block ([16] Issurin 2008, Level 5). The [17] Matveyev 1981 fundamentals of sports training placed the same on the multi-year periodization side: adaptation is a long-horizon process; the single variable being tested is held across a meaningful block ([17] Matveyev 1981, Level 5).

Duration first: the safest dial

Duration is the default dial — the variable to change when in doubt. The [3] Seiler 2010 polarised-training framework placed duration on the endurance side: low-intensity duration is the largest component of endurance-training load, and the first dial to grow in any endurance program is duration ([3] Seiler 2010, Level 1a/2a). The [26] Pollock 1978 recommended-quantity-and-quality review placed the same on the foundational-dose-response side: duration is the most reliable dose dimension in early progression; the canonical first move is to add minutes ([26] Pollock 1978, Level 5).

The [11] Garber 2011 ACSM position stand placed duration on the canonical-progression side: dose-response evidence supports a progressive increase in duration as the first lever in cardiorespiratory progression ([11] Garber 2011, Level 5). The [27] Astrand 2003 textbook of work physiology placed the same on the work-physiology side: low-intensity duration drives the largest portion of the chronic-load signal; the aerobic base is built across weeks of accumulating duration ([27] Astrand 2003, Level 5). The [28] Murtagh 2018 rowing-specific load-management review in IJSPP placed the same on the rowing side: the largest dial in rowing-specific load is duration, and the first move is to grow duration before pace ([28] Murtagh 2018, Level 1a).

The operational read: when in doubt, hold pace and recovery, add one to two minutes of duration. The [16] Issurin 2008 block-periodization review placed this on the block-organization side: within a focused block, duration is the load-bearing dial; pace is held constant ([16] Issurin 2008, Level 5). The [19] Kraemer 2004 progression-prescription framework placed the same on the prescription side: incremental dose-response in cardiorespiratory training begins with duration ([19] Kraemer 2004, Level 5). The [24] Burke 2011 carbohydrate-intake framework placed the same on the metabolic side: a rower whose duration is growing needs the carbohydrate support to back it up; the variable change is being read in two places at once, and the trend on the ergometer should match the trend in fuel tolerance ([24] Burke 2011, Level 5).

Repetitions next: when to add one

Repetitions are the next dial. The [30] ACSM 2009 progression-models position stand placed repetitions on the canonical-dose-response side: dose-response work identifies frequency as a primary variable, and repetitions within a session are the load-bearing realization of frequency ([30] ACSM 2009, Level 5). The [19] Kraemer 2004 progression-prescription framework placed the same on the prescription side: progression in cardiorespiratory work moves from duration to repetitions before pace ([19] Kraemer 2004, Level 5).

The [16] Issurin 2008 block-periodization review placed repetitions on the block-organization side: within a focused block, the repetition scheme is the structural identity of the block; across blocks, the repetition scheme shifts ([16] Issurin 2008, Level 5). The [27] Astrand 2003 textbook of work physiology placed the same on the work-physiology side: the repeat structure determines whether the session is aerobic-endurance work or threshold work; adding one repetition is a real load increase when pace and recovery are held ([27] Astrand 2003, Level 5). The [20] Rietjens 2001 rowing-specific periodization paper in IJSPP placed the same on the rowing side: rowing progression moves from duration to repetition scheme before pace ([20] Rietjens 2001, Level 1b/2b).

The operational read: when duration has stopped producing adaptation, hold pace and duration and add one repetition. The [28] Murtagh 2018 rowing-specific load-management review placed this on the rowing-specific side: the second dial in rowing-specific progression is repetitions, and the first move when duration plateaus is to add one repetition ([28] Murtagh 2018, Level 1a). The [11] Garber 2011 ACSM position stand placed the same on the canonical-progression side: dose-response curves show a stepwise progression from frequency to duration to intensity ([11] Garber 2011, Level 5).

The honest read: adding one repetition is a smaller load increase than tightening recovery or pushing pace. The [16] Issurin 2008 block-periodization review placed the same on the block-organization side: within a focused block, repetition changes are the load-bearing dial; pace is held ([16] Issurin 2008, Level 5). The [3] Seiler 2010 polarised-training framework placed the same on the endurance side: endurance progression is built across many sessions of adding minutes and repetitions, not pace ([3] Seiler 2010, Level 1a/2a).

Recovery: tightening the gap

Recovery is the third dial. The [4] Halson 2014 training-load monitoring review placed recovery on the multi-modal-signal side: tightening recovery is a real load increase, and the multi-modal signal catches the change ([4] Halson 2014, Level 5). The [1] Foster 2001 session-RPE method placed the same on the load-monitoring side: sRPE creeps when recovery tightens, and the load calculation catches the change ([1] Foster 2001, Level 5).

The [6] Impellizzeri 2019 injury-and-illness-prevention review placed recovery on the chronic-load side: chronic load is sensitive to recovery, and tightening recovery without an explicit recovery dial changes the chronic-load trajectory ([6] Impellizzeri 2019, Level 1a). The [16] Issurin 2008 block-periodization review placed the same on the block-organization side: across a focused block, recovery is held constant; within the block, recovery is the third dial to tighten, not the first ([16] Issurin 2008, Level 5).

The operational read: when duration and repetitions have stopped producing adaptation, hold pace, hold duration, and tighten recovery. The [28] Murtagh 2018 rowing-specific load-management review placed this on the rowing-specific side: the third dial in rowing-specific progression is recovery, and tightening recovery is a real load increase the rower feels as a rise in sRPE before they see it on the ergometer ([28] Murtagh 2018, Level 1a). The [11] Garber 2011 ACSM position stand placed the same on the canonical-progression side: dose-response curves identify recovery as a derivative variable; progression in recovery is built on top of stable duration and repetitions ([11] Garber 2011, Level 5).

The honest read for the rower: tightening recovery is the dial most often confounded with pace. The [22] Sawka 2007 ACSM exertional-heat-illness position stand placed the same on the safety side: tightening recovery in heat raises heat-illness risk before it raises adaptation; pace and recovery are the safety-sensitive dials and should be touched last ([22] Sawka 2007, Level 5). The [4] Halson 2014 training-load monitoring review placed the same on the multi-modal-signal side: HR + sRPE + duration drift together catch the recovery tightening before the rower feels it ([4] Halson 2014, Level 5).

Pace: the last dial to touch

Pace is the last dial to touch — the most sensitive, the easiest to get wrong. The [25] Swain 2006 vigorous-vs-moderate review placed pace on the dose-response side: pace is the most intensity-sensitive dial, and the dose-response curve for pace is steeper than the dose-response curve for duration ([25] Swain 2006, Level 5). The [3] Seiler 2010 polarised-training framework placed the same on the endurance side: pace is held steady across most of the endurance-training load; pace is the variable touched only in focused blocks ([3] Seiler 2010, Level 1a/2a).

The [13] Tanaka 2001 HRmax formula in JACC placed pace on the rate-cap side: pace is constrained by HR; the rate-cap is rate-band-specific, and pace is the variable that walks the rower up the rate-band ([13] Tanaka 2001, Level 1b). The [14] Karvonen 1957 HR-reserve formula placed the same on the rate-band side: HR reserve is the linear transform that turns absolute HR into a rate-band; pace is the variable that drives HR reserve ([14] Karvonen 1957, Level 5). The [22] Sawka 2007 ACSM exertional-heat-illness position stand placed the same on the safety side: pace is the safety-sensitive dial; touching pace in heat or under fatigue raises heat-illness risk ([22] Sawka 2007, Level 5).

The operational read: pace is the dial to touch only when duration, repetitions, and recovery have stopped producing adaptation, and only under controlled conditions. The [28] Murtagh 2018 rowing-specific load-management review placed this on the rowing-specific side: the last dial in rowing-specific progression is pace, and pace is held until the other three dials have run their course ([28] Murtagh 2018, Level 1a). The [16] Issurin 2008 block-periodization review placed the same on the block-organization side: across a focused block, pace is held constant; the block ends and a new block begins with a pace shift ([16] Issurin 2008, Level 5).

The honest read for the rower: pace is the dial most often confounded with the others. The [1] Foster 2001 session-RPE method placed the same on the load-monitoring side: sRPE creeps when pace creeps, and the load calculation catches the change before the rower notices ([1] Foster 2001, Level 5). The [5] Borg 1982 CR-10 RPE scale placed the same on the perceived-exertion side: RPE is the rower's second-channel read; a change in RPE without a change in objective load tells the rower the dial that moved was pace ([5] Borg 1982, Level 5). The [4] Halson 2014 training-load monitoring review placed the same on the multi-modal-signal side: HR + sRPE + duration drift together catch the pace creep before the rower attributes it to a different cause ([4] Halson 2014, Level 5).

The AI coach's role

The AI coach chooses which dial to change based on the rower's trend. The [7] Plews 2018 evaluating-adaptation paper in IJSPP placed the AI-coach role on the HRV side: HRV is the diagnostic for adaptation, and the AI coach holds other variables constant while testing the dial that needs adjustment ([7] Plews 2018, Level 1b). The [8] Vesterinen 2016 HRV-guided field trial in MSSE placed the same on the field-trial side: HRV-guided prescription works precisely because the algorithm holds other variables constant while testing the dial that needs adjustment ([8] Vesterinen 2016, Level 1b/2b).

The [4] Halson 2014 training-load monitoring review placed the AI-coach role on the multi-modal-signal side: the multi-modal signal is the diagnostic, and the AI coach uses the diagnostic to choose the dial ([4] Halson 2014, Level 5). The [1] Foster 2001 session-RPE method placed the same on the load-monitoring side: sRPE is one input to the multi-modal signal, and the AI coach uses sRPE across a multi-week window to choose the dial ([1] Foster 2001, Level 5). The [6] Impellizzeri 2019 injury-and-illness-prevention review placed the same on the chronic-load side: chronic load is the safety input, and the AI coach uses chronic load to choose the dial ([6] Impellizzeri 2019, Level 1a).

The operational read: the AI coach chooses duration first, repetitions next, recovery next, and pace last. The [11] Garber 2011 ACSM position stand placed the same on the canonical-progression side: the canonical dose-response curve shows a stepwise progression from frequency to duration to intensity, and the AI coach follows the canonical sequence ([11] Garber 2011, Level 5). The [16] Issurin 2008 block-periodization review placed the same on the block-organization side: within a focused block, the dial is held constant; across blocks, the AI coach shifts the dial ([16] Issurin 2008, Level 5). The [18] Bompa 2019 periodization textbook placed the same on the textbook side: periodization follows a sequential logic, and the AI coach implements the sequential logic rower-by-rower ([18] Bompa 2019, Level 5).

The honest read for the rower: the AI coach is the diagnostic, and the rower is the executor. The [9] Mann 2014 high-vs-low-responder study placed the same on the individual-variation side: rower-to-rower response varies; the AI coach individualises the diagnostic, but the rower holds the dial constant for the watch period ([9] Mann 2014, Level 2b). The [10] Buchheit 2014 HR-monitoring review placed the same on the trend-monitoring side: HR-derived fatigue markers are sensitive to multiple variables; the AI coach holds other variables constant to keep the trend clean ([10] Buchheit 2014, Level 5). The [21] Platonov 2006 multi-year periodization framework placed the same on the long-horizon side: the AI coach operates on a multi-week window inside a multi-year horizon, and the long-horizon view is what makes the one-variable-at-a-time rule sustainable ([21] Platonov 2006, Level 5).

The AI coach's diagnostic logic. The [7] Plews 2018 evaluating-adaptation paper placed the diagnostic on the HRV side: the AI coach reads HRV as the primary signal, with sRPE and duration drift as the secondary signals; the rower reports sRPE, the AI coach reads HRV, and the multi-modal signal is the diagnostic ([7] Plews 2018, Level 1b). The [8] Vesterinen 2016 HRV-guided field trial placed the same on the field-trial side: the AI coach adjusts the prescription based on the rolling HRV trend, and the rower reports the watch-period read at the end of the watch window ([8] Vesterinen 2016, Level 1b/2b). The [6] Impellizzeri 2019 injury-and-illness-prevention review placed the same on the chronic-load side: the AI coach computes chronic-vs-acute load in real time and adjusts the next session before the rower notices fatigue ([6] Impellizzeri 2019, Level 1a).

The AI coach's protective logic. The [11] Garber 2011 ACSM position stand placed the protective logic on the canonical-progression side: the AI coach holds the rower to the dose-response curve, and the rower who wants to push past the curve is held back by the AI coach's read of the trend ([11] Garber 2011, Level 5). The [22] Sawka 2007 ACSM exertional-heat-illness position stand placed the same on the safety side: the AI coach can refuse to progress on pace during a heat event, even when the rower is asking to push; the safety read takes precedence ([22] Sawka 2007, Level 5). The [12] Pescatello 2021 ACSM Guidelines reached the same conclusion from the clinical-prescription side: the AI coach holds the rower to clinical-prescription evidence, and the rower's urge to break the rule is filtered through the evidence base ([12] Pescatello 2021, Level 5).

The AI coach's failure modes. The AI coach is not omniscient, and the watch period is not a guarantee. The [9] Mann 2014 high-vs-low-responder study placed the AI-coach failure case on the individual-variation side: a high-responder can be under-progressed if the AI coach holds the dial too long, and a low-responder can be over-progressed if the AI coach advances too quickly ([9] Mann 2014, Level 2b). The [4] Halson 2014 training-load monitoring review placed the same on the multi-modal-signal side: the multi-modal signal is incomplete without the rower's reported state, and the AI coach who relies only on objective markers misses the rower's experience ([4] Halson 2014, Level 5). The honest read: the AI coach is the diagnostic, but the rower's reported state is the second half of the diagnostic; the rower who keeps the watch-period log honest is the rower whose AI-coach read is accurate.

Practical reading: how to apply the rule

The practical read follows the canonical dose-response curve. The [11] Garber 2011 ACSM position stand placed the sequence on the canonical-progression side: progression in cardiorespiratory fitness follows frequency, then duration, then intensity ([11] Garber 2011, Level 5). The [12] Pescatello 2021 ACSM Guidelines placed the same on the clinical-prescription side: prescription is anchored to dose-response evidence, and the dose-response evidence is built on one-variable-at-a-time trials ([12] Pescatello 2021, Level 5).

The [30] ACSM 2009 progression-models position stand placed the practical sequence on the canonical-dose-response side: progression in cardiorespiratory work moves from frequency to duration to intensity; repetitions are a structural realization of frequency; recovery is a derivative variable ([30] ACSM 2009, Level 5). The [28] Murtagh 2018 rowing-specific load-management review placed the same on the rowing-specific side: the rowing-specific sequence is duration first, repetitions next, recovery next, pace last ([28] Murtagh 2018, Level 1a). The [20] Rietjens 2001 rowing-specific periodization paper placed the same on the rowing side: rowing progression moves from duration to repetition scheme before pace ([20] Rietjens 2001, Level 1b/2b).

The practical read:

1. Pick the dial. Start with duration. The [3] Seiler 2010 polarised-training framework and the [11] Garber 2011 ACSM position stand placed this on the canonical-progression side: duration first ([3] Seiler 2010, Level 1a/2a; [11] Garber 2011, Level 5).
2. Hold the others. Hold pace, repetitions, and recovery constant. The [15] Kiely 2012 periodization critique placed this on the evidence side: holding other variables constant is the only way to read the result ([15] Kiely 2012, Level 5).
3. Make the change. Add one to two minutes of duration, or one repetition, or one second of tightened recovery, or one second of pace. The [1] Foster 2001 session-RPE method placed this on the load-monitoring side: the change should be detectable on the multi-modal signal ([1] Foster 2001, Level 5).
4. Watch the trend for two weeks. The [8] Vesterinen 2016 HRV-guided field trial placed this on the adaptive-prescription side: the trend takes weeks to settle ([8] Vesterinen 2016, Level 1b/2b).
5. Read the multi-modal signal. HR + sRPE + duration drift. The [4] Halson 2014 training-load monitoring review placed this on the multi-modal-signal side ([4] Halson 2014, Level 5).
6. If the trend is positive, keep the change. The [11] Garber 2011 ACSM position stand placed this on the canonical-progression side: positive trend means the dial is producing adaptation ([11] Garber 2011, Level 5).
7. If the trend is flat, wait one more week. The [9] Mann 2014 high-vs-low-responder study placed this on the individual-variation side: high-responders adapt faster, low-responders slower ([9] Mann 2014, Level 2b).
8. If the trend is negative, revert the change. The [6] Impellizzeri 2019 injury-and-illness-prevention review placed this on the chronic-load side: a negative trend on chronic load is a warning ([6] Impellizzeri 2019, Level 1a).
9. Move to the next dial only when the current dial has plateaued. The [16] Issurin 2008 block-periodization review placed this on the block-organization side: a focused block ends when the dial has run its course ([16] Issurin 2008, Level 5).
10. Never touch two dials in the same session. The [15] Kiely 2012 periodization critique and the [4] Halson 2014 review placed this on the evidence and multi-modal-signal sides: two dials produce an uninterpretable signal ([15] Kiely 2012, Level 5; [4] Halson 2014, Level 5).

Edge case: returning to training after a break

The rule is different when the rower is returning after a break. The [16] Issurin 2008 block-periodization review placed the return-to-training case on the block-organization side: a return is itself a focused block, but the dial is "duration" at a low starting point and the AI coach holds pace at a conservative cap ([16] Issurin 2008, Level 5). The [28] Murtagh 2018 rowing-specific load-management review placed the same on the rowing-specific side: rowing-specific return-to-training begins with duration at 50-60% of the prior baseline and climbs across the watch period ([28] Murtagh 2018, Level 1a). The honest read: a returning rower is a low-responder by default, and the watch period should be longer than the standard two weeks.

The [22] Sawka 2007 ACSM exertional-heat-illness position stand placed the return-to-training case on the safety side: a rower returning after a heat-related break is at higher heat-illness risk, and the pace dial should be held below the prior rate-cap ([22] Sawka 2007, Level 5). The [11] Garber 2011 ACSM position stand placed the same on the canonical-progression side: dose-response curves for returning rowers are slower than for continuous-training rowers, and the watch period should be calibrated to the returning-rower curve ([11] Garber 2011, Level 5). The [9] Mann 2014 high-vs-low-responder study placed the same on the individual-variation side: a returning rower may have lost adaptation in the break, and the response profile may have shifted; the AI coach recalibrates the watch period based on the rolling trend ([9] Mann 2014, Level 2b).

The [1] Foster 2001 session-RPE method placed the return-to-training case on the load-monitoring side: sRPE is the load-bearing signal, and a returning rower's sRPE is naturally higher at the same absolute load; the watch period on duration should account for the elevated sRPE ([1] Foster 2001, Level 5). The [4] Halson 2014 training-load monitoring review reached the same conclusion from the multi-modal-signal side: HR + sRPE + duration drift together tell the AI coach whether the returning rower is ready to progress ([4] Halson 2014, Level 5). The [6] Impellizzeri 2019 injury-and-illness-prevention review placed the same on the chronic-load side: chronic load drops during the break, and the return is itself a re-build of chronic load; the watch period is a fresh chronic-load build ([6] Impellizzeri 2019, Level 1a).

The honest read for the returning rower: the rule still applies, but the starting point and the watch period are different. The [16] Issurin 2008 block-periodization review placed the same on the block-organization side: a return is a focused block with a low starting duration and a conservative pace cap, and the watch period is longer than the standard two weeks ([16] Issurin 2008, Level 5). The [18] Bompa 2019 periodization textbook placed the same on the textbook side: a return-to-training is a sequential block, and the AI coach implements the sequential block rower-by-rower ([18] Bompa 2019, Level 5). The rower who returns with the rule intact — duration first, pace last, watch period calibrated — is the rower who rebuilds without re-injury.

Edge case: plateaus and when the dial has run its course

A plateau is the signal that the dial has run its course. The [16] Issurin 2008 block-periodization review placed the plateau on the block-organization side: a focused block ends when the dial has run its course, and the next block starts with a new dial ([16] Issurin 2008, Level 5). The [11] Garber 2011 ACSM position stand placed the same on the canonical-progression side: dose-response curves plateau when the canonical dose has been reached; the rower who has plateaued on duration should move to repetitions ([11] Garber 2011, Level 5). The [9] Mann 2014 high-vs-low-responder study placed the same on the individual-variation side: a plateau on duration for one rower may not be a plateau for another; the AI coach reads the multi-modal signal to determine whether the plateau is real ([9] Mann 2014, Level 2b).

The [8] Vesterinen 2016 HRV-guided field trial placed the plateau-detection case on the adaptive-prescription side: HRV-guided prescription detects plateau by the rolling HRV trend, not by a fixed week count; a rower whose HRV is still adapting is not at a plateau ([8] Vesterinen 2016, Level 1b/2b). The [7] Plews 2018 evaluating-adaptation paper reached the same conclusion from the HRV side: HRV-based fatigue markers detect plateau by the rolling trend; a rower whose HRV is flat but sRPE is climbing is at a plateau on adaptation, not at a plateau on the watch ([7] Plews 2018, Level 1b). The honest read: a plateau on the watch is a signal to move to the next dial; a plateau on the multi-modal signal is a signal to wait or revert.

The [3] Seiler 2010 polarised-training framework placed the plateau case on the endurance side: a rower who has plateaued on duration at the polarised level has reached the canonical endurance dose; the next move is intensity, not more duration ([3] Seiler 2010, Level 1a/2a). The [27] Astrand 2003 textbook of work physiology placed the same on the work-physiology side: a rower who has plateaued on pace at the rate-cap has reached the rate-band ceiling; the next move is recovery, not more pace ([27] Astrand 2003, Level 5). The [28] Murtagh 2018 rowing-specific load-management review reached the same conclusion from the rowing-specific side: a rower who has plateaued on all four dials at the rowing-specific ceiling is at the sport-specific ceiling, and the next move is a deload or a transition block ([28] Murtagh 2018, Level 1a).

The honest read for the rower: a plateau is a feature, not a bug. The [16] Issurin 2008 block-periodization review placed the plateau on the block-organization side: a plateau signals the end of a focused block, and the end of a block is the right time to shift dials ([16] Issurin 2008, Level 5). The [18] Bompa 2019 periodization textbook reached the same conclusion from the textbook side: a plateau on one dial is the trigger for the next block, and the next block begins with the dial that has the highest unexploited potential ([18] Bompa 2019, Level 5). The [21] Platonov 2006 multi-year periodization framework placed the same on the long-horizon side: a plateau on a multi-week block is a unit of progression, and the multi-year plan stacks units across years ([21] Platonov 2006, Level 5).

Edge case: the rate-cap and the HR-reserve calculation

The rate-cap and the HR-reserve calculation are useful when pace is the dial being tested. The [13] Tanaka 2001 HRmax formula in JACC placed the rate-cap on the HRmax side: HRmax ≈ 208 − (0.7 × age), and the rate-cap sits below HRmax based on the rower's training age ([13] Tanaka 2001, Level 1b). The [14] Karvonen 1957 HR-reserve formula placed the same on the HR-reserve side: HR reserve = HRmax − HRrest, and the rate-cap sits at a percentage of HR reserve based on the rower's target intensity ([14] Karvonen 1957, Level 5). The [25] Swain 2006 vigorous-vs-moderate review placed the same on the dose-response side: the rate-cap is the upper bound on pace, and the rower who touches pace past the rate-cap is reading the wrong dial ([25] Swain 2006, Level 5).

The [22] Sawka 2007 ACSM exertional-heat-illness position stand placed the rate-cap on the safety side: in heat, the rate-cap drops because HRmax is depressed and HR reserve is compressed; a rower who touches pace in heat without recalculating the rate-cap is at heat-illness risk ([22] Sawka 2007, Level 5). The [4] Halson 2014 training-load monitoring review reached the same conclusion from the multi-modal-signal side: HR + sRPE + duration drift together tell the AI coach whether the rate-cap is being respected ([4] Halson 2014, Level 5). The [27] Astrand 2003 textbook of work physiology placed the same on the work-physiology side: the rate-cap is the ceiling on intensity, and the watch period on pace respects the ceiling ([27] Astrand 2003, Level 5).

The honest read for the rower: the rate-cap is a tool for pace, not for duration. The [28] Murtagh 2018 rowing-specific load-management review placed the rate-cap use-case on the rowing-specific side: the rate-cap applies to pace, not to duration or repetitions; the rower who uses the rate-cap during a duration block is testing the wrong dial ([28] Murtagh 2018, Level 1a). The [16] Issurin 2008 block-periodization review reached the same conclusion from the block-organization side: within a focused block, the rate-cap applies only to the dial being tested ([16] Issurin 2008, Level 5). The [12] Pescatello 2021 ACSM Guidelines placed the same on the clinical-prescription side: prescription anchors rate-cap calculations to dose-response evidence, and the rower who uses the rate-cap outside the dose-response curve is operating outside the evidence base ([12] Pescatello 2021, Level 5).

Common confounds: the variables outside the four dials

The four dials are not the only variables that can confound the trend. The [23] Helms 2014 nutrition-and-training framework placed nutrition on the parallel-side: the same one-variable-at-a-time logic applies to nutrition, and a rower whose carbohydrate or protein intake is changing at the same time as the training dial is changing is touching two variables at once ([23] Helms 2014, Level 5). The [24] Burke 2011 carbohydrate-intake framework reached the same conclusion from the carbohydrate side: a rower whose carbohydrate availability is changing across the watch period is confounding the training-side signal with a nutrition-side signal ([24] Burke 2011, Level 5). The honest read: when a training dial is being tested, the nutrition dial should be held for at least the two-week watch period.

Sleep is another confound. The [4] Halson 2014 training-load monitoring review placed sleep on the recovery-side: HR + sRPE + duration drift catch sleep changes, but only when sleep is on the multi-modal signal; a rower whose sleep is varying across the watch period produces an ambiguous trend ([4] Halson 2014, Level 5). The [9] Mann 2014 high-vs-low-responder study placed the same on the individual-variation side: a rower whose sleep drops by an hour during the watch period is producing a confound that the rower may not notice ([9] Mann 2014, Level 2b). The honest read: when a training dial is being tested, sleep should be reported alongside sRPE; if sleep varies by more than an hour across the watch period, the watch period should be extended.

Equipment is another confound. The [16] Issurin 2008 block-periodization review placed equipment on the block-organization side: a rower who changes drag factor, foot-stretchers, or handle grip during the watch period has introduced a second variable ([16] Issurin 2008, Level 5). The [28] Murtagh 2018 rowing-specific load-management review placed the same on the rowing-specific side: rowing-specific progression holds equipment constant across the focused block ([28] Murtagh 2018, Level 1a). The honest read: equipment changes should sit outside the watch period; if equipment must change (a new ergometer, a different drag setting), the watch period resets.

Technique is the most subtle confound. The [19] Kraemer 2004 progression-prescription framework placed technique on the prescription side: a rower whose technique is changing during the watch period is confounding the training-side signal with a technique-side signal; the rower who changes the stroke pattern cannot tell whether the trend is from the dial change or from the technique change ([19] Kraemer 2004, Level 5). The [3] Seiler 2010 polarised-training framework reached the same conclusion from the endurance side: technique changes should sit outside the watch period; the technique block and the volume block are sequential, not parallel ([3] Seiler 2010, Level 1a/2a). The honest read: technique work should be done in a separate block; the watch period is for one dial at a time, and the dial is a training dial.

A worked example: applying the rule across four weeks

The worked example walks through a four-week block. The [16] Issurin 2008 block-periodization review placed the focused block on the operational side: a focused block is the unit of progression, and the block has a clear dial, a clear duration, and a clear read ([16] Issurin 2008, Level 5). The [11] Garber 2011 ACSM position stand placed the same on the canonical-progression side: incremental dose-response is the cornerstone of cardiorespiratory prescription; a four-week block is a reasonable read window for a single dial ([11] Garber 2011, Level 5). The [8] Vesterinen 2016 HRV-guided field trial placed the adaptive-prescription side: HRV-guided adjustment works on a multi-week horizon, and the four-week block is the minimum window for adaptation ([8] Vesterinen 2016, Level 1b/2b).

Week 0 — baseline. The rower runs the current session three times across one week, with all four dials held constant. The [4] Halson 2014 training-load monitoring review placed the baseline on the multi-modal-signal side: HR + sRPE + duration drift across three sessions is the baseline; the AI coach uses the baseline to detect the trend ([4] Halson 2014, Level 5). The [1] Foster 2001 session-RPE method placed the same on the load-monitoring side: sRPE × duration across the baseline is the load-bearing number ([1] Foster 2001, Level 5).

Week 1 — Durability block. The dial is duration. The rower adds one minute to each session, holds pace, repetitions, and recovery. The [3] Seiler 2010 polarised-training framework placed duration on the largest-load side: low-intensity duration is the largest dial in endurance programming ([3] Seiler 2010, Level 1a/2a). The [26] Pollock 1978 recommended-quantity-and-quality review placed the same on the foundational-dose-response side: duration is the most reliable first dial ([26] Pollock 1978, Level 5).

Week 2 — durability read. The AI coach compares the week-1 trend against the baseline. The [8] Vesterinen 2016 HRV-guided field trial placed the read on the adaptive-prescription side: HRV + sRPE + duration drift across the two-week window tells the coach whether the dial change is producing adaptation ([8] Vesterinen 2016, Level 1b/2b). The [9] Mann 2014 high-vs-low-responder study placed the same on the individual-variation side: rower-to-rower response varies; some rowers adapt in week 2, others in week 3 ([9] Mann 2014, Level 2b).

Week 3 — repetition block. The dial is repetitions. The rower holds pace, duration, and recovery, adds one repetition. The [19] Kraemer 2004 progression-prescription framework placed repetitions on the prescription side: progression in cardiorespiratory work moves from duration to repetitions before pace ([19] Kraemer 2004, Level 5). The [16] Issurin 2008 block-periodization review placed the same on the block-organization side: within a focused block, the dial shifts; across blocks, the dial holds ([16] Issurin 2008, Level 5).

Week 4 — repetition read. The AI coach compares the week-3 trend against the week-1 trend. The [11] Garber 2011 ACSM position stand placed the read on the canonical-progression side: positive trend on the new dial means adaptation is occurring ([11] Garber 2011, Level 5). The [6] Impellizzeri 2019 injury-and-illness-prevention review placed the same on the safety side: chronic-vs-acute load should stay within the safe band; a negative trend is a warning ([6] Impellizzeri 2019, Level 1a).

The four-week block ends with the rower on a higher total load than the baseline, with each adaptation attributable to a specific dial change. The [18] Bompa 2019 periodization textbook placed the sequential-block logic on the textbook side: sequential blocks are the periodization sequence, and the rower who follows the sequence can read each adaptation ([18] Bompa 2019, Level 5). The [17] Matveyev 1981 fundamentals of sports training placed the same on the multi-year-development side: the four-week block is a unit of progression, and the long-horizon view stacks units into a multi-year plan ([17] Matveyev 1981, Level 5).

When to abandon the rule: recovery weeks, taper, illness, deload

The one-variable-at-a-time rule has exceptions. The [16] Issurin 2008 block-periodization review placed the recovery week on the block-organization side: a recovery week is built into the block, and the recovery week hits all four dials at once ([16] Issurin 2008, Level 5). The [18] Bompa 2019 periodization textbook reached the same conclusion from the textbook side: a recovery week is the load reset, and the load reset hits duration, repetitions, recovery, and pace simultaneously ([18] Bompa 2019, Level 5). The honest read: a recovery week is the right place to abandon the rule, because the goal of the recovery week is to reset the load, not to read a single dial.

The taper is another exception. The [16] Issurin 2008 block-periodization review placed the taper on the focused-block side: a taper reduces load across all four dials, and the taper is the bridge between the build phase and the test phase ([16] Issurin 2008, Level 5). The [21] Platonov 2006 multi-year periodization framework placed the same on the long-horizon side: a taper is a multi-week event, and the taper itself is not a one-variable-at-a-time read ([21] Platonov 2006, Level 5). The honest read: a taper is the right place to abandon the rule, because the taper is a structured reduction, not a single-dial test.

Illness is a different exception. The [22] Sawka 2007 ACSM exertional-heat-illness position stand placed illness on the safety side: a rower who is ill should not be progressing on any dial, and the return to training is a multi-week process ([22] Sawka 2007, Level 5). The [6] Impellizzeri 2019 injury-and-illness-prevention review reached the same conclusion from the chronic-load side: chronic load drops during illness, and the return is itself a one-variable-at-a-time read across multiple dials ([6] Impellizzeri 2019, Level 1a). The honest read: illness is the right place to abandon the rule entirely, and the return-to-training is a fresh start with the watch period reset.

A deload week is a planned exception. The [16] Issurin 2008 block-periodization review placed the deload on the block-organization side: a deload is built into the focused block, and the deload hits all four dials at once ([16] Issurin 2008, Level 5). The [11] Garber 2011 ACSM position stand placed the same on the canonical-progression side: dose-response curves include deload as part of the cycle; deload is the recovery that makes the next build possible ([11] Garber 2011, Level 5). The honest read: a deload week is the right place to abandon the rule temporarily, because the goal of the deload is to reset the chronic-load trajectory, not to test a single dial.

The honest read for the rower: the rule is for progression, not for recovery. The [18] Bompa 2019 periodization textbook placed progression on the build phase, and recovery on the recovery phase: the rule applies to the build phase, and the recovery phase is the structured exception ([18] Bompa 2019, Level 5). The [16] Issurin 2008 block-periodization review reached the same conclusion from the block-organization side: blocks alternate between build and recovery, and the rule applies to the build blocks ([16] Issurin 2008, Level 5). The rower who follows the rule during build and abandons the rule during recovery is the rower who can sustain progression across years.

Edge case: what counts as a "variable"

Not every change is a "variable" in the rule's sense. The [2] Banister 1975 TRIMP framework placed the variable-definition on the load-product side: TRIMP = duration × intensity, and the variables that change TRIMP are the variables the rule applies to ([2] Banister 1975, Level 5). The [1] Foster 2001 session-RPE method placed the same on the load-monitoring side: sRPE × duration is the load, and the variables that change sRPE or duration are the rule's variables ([1] Foster 2001, Level 5). The honest read: the rule applies to the inputs that change the load calculation; inputs that do not change the load are not variables in the rule's sense.

The [23] Helms 2014 nutrition-and-training framework placed the variable-definition on the parallel-side: a nutrition change that does not affect the load calculation is not a variable; a nutrition change that does affect the load calculation is a variable ([23] Helms 2014, Level 5). The [24] Burke 2011 carbohydrate-intake framework reached the same conclusion from the carbohydrate side: carbohydrate availability can change the load by changing the metabolic ceiling, but a small carbohydrate change that does not change the metabolic ceiling is not a variable ([24] Burke 2011, Level 5). The honest read: a rower who eats a slightly larger breakfast before a session is not touching a variable, but a rower who shifts from low-carb to high-carb across the watch period is touching a variable.

The [16] Issurin 2008 block-periodization review placed the variable-definition on the block-organization side: within a focused block, the variable being tested is the dial that the AI coach is changing; variables outside the dial are held constant ([16] Issurin 2008, Level 5). The [28] Murtagh 2018 rowing-specific load-management review placed the same on the rowing-specific side: rowing-specific variables are duration, repetitions, recovery, pace; non-rowing-specific changes (sleep, nutrition, equipment) are not variables in the rule's sense unless they change the load ([28] Murtagh 2018, Level 1a). The honest read: the rule applies to the dial that changes the load; outside-the-dial changes are tracked separately and are not the rule's variables.

The [4] Halson 2014 training-load monitoring review placed the variable-definition on the multi-modal-signal side: the multi-modal signal catches changes in the load-bearing variables and changes in the outside-the-dial variables; the AI coach separates the two and reads only the load-bearing variables ([4] Halson 2014, Level 5). The [10] Buchheit 2014 HR-monitoring review reached the same conclusion from the HR-side: HR-derived fatigue markers are sensitive to both load-bearing variables and outside-the-dial variables; the AI coach separates the channels ([10] Buchheit 2014, Level 5). The rower who can separate the channels is the rower who can read the trend cleanly; the rower who conflates the channels produces a noisy trend.

Edge case: rowers who train multiple sports

The rule adapts for rowers who train multiple sports. The [19] Kraemer 2004 progression-prescription framework placed the multi-sport case on the prescription side: a rower who also cycles or runs has more than one activity contributing to chronic load, and the dial being tested must account for the cross-sport load ([19] Kraemer 2004, Level 5). The [6] Impellizzeri 2019 injury-and-illness-prevention review placed the same on the chronic-load side: chronic load is the sum of all sport-loads, and a rower who adds a cycling session during the watch period is touching a second variable ([6] Impellizzeri 2019, Level 1a). The honest read: the watch period is for the ergometer dial; cross-sport load is held constant across the watch period.

The [20] Rietjens 2001 rowing-specific periodization paper placed the multi-sport case on the rowing side: rowing-specific periodization is the rowing dial, but cross-training is a separate dial, and the watch period on the rowing dial holds cross-training constant ([20] Rietjens 2001, Level 1b/2b). The [3] Seiler 2010 polarised-training framework placed the same on the endurance side: cross-training is itself a dial, and the watch period on one sport's dial holds the cross-training dial constant ([3] Seiler 2010, Level 1a/2a). The [28] Murtagh 2018 rowing-specific load-management review reached the same conclusion from the rowing-specific side: a rowing watch period holds cross-training constant, and the cross-training dial is tested in a separate block ([28] Murtagh 2018, Level 1a).

The [4] Halson 2014 training-load monitoring review placed the multi-sport multi-modal signal on the diagnostic side: HR + sRPE + duration drift is the rowing signal, but cross-training produces its own multi-modal signal, and the AI coach separates the two ([4] Halson 2014, Level 5). The [1] Foster 2001 session-RPE method placed the same on the load-monitoring side: sRPE is sport-specific, and a rowing sRPE is different from a cycling sRPE; the watch period uses rowing sRPE ([1] Foster 2001, Level 5). The honest read: a rower who trains multiple sports has more channels to track, and the watch period is for one sport at a time.

Limitations

The one-variable-at-a-time rule has failure modes. The [9] Mann 2014 high-vs-low-responder study placed the rower-to-rower variation on the individual side: response to a single variable change varies by an order of magnitude; the same dial change can produce a fast adaptation in one rower and a slow one in another ([9] Mann 2014, Level 2b). The [10] Buchheit 2014 HR-monitoring review placed the same on the trend-monitoring side: HR-derived fatigue markers fluctuate day-to-day; the trend is reliable, but the single-session read is noisy ([10] Buchheit 2014, Level 5).

The [4] Halson 2014 training-load monitoring review placed the same on the multi-modal-signal side: single markers misfire; the constellation of HR + sRPE + duration drift is the load-bearing signal, but the constellation is incomplete without the rower's reported state ([4] Halson 2014, Level 5). The [22] Sawka 2007 ACSM exertional-heat-illness position stand placed the same on the safety side: pace is the safety-sensitive dial; touching pace in heat or under fatigue raises heat-illness risk; the two-week watch period does not protect against acute heat events ([22] Sawka 2007, Level 5).

The [21] Platonov 2006 multi-year periodization framework placed the same on the long-horizon side: the one-variable-at-a-time rule works on a multi-week window inside a multi-year horizon; the long-horizon view is what makes the rule sustainable, but the rule is not the only input to a multi-year program ([21] Platonov 2006, Level 5). The [17] Matveyev 1981 fundamentals of sports training placed the same on the multi-year-development side: adaptation is a long-horizon process; the single-variable rule is a constraint on a much larger system ([17] Matveyev 1981, Level 5).

The [23] Helms 2014 nutrition-and-training framework placed the same on the parallel-side: the same one-variable-at-a-time logic applies to nutrition, and the rower's training and nutrition interact; touching a training dial while a nutrition dial is also moving produces the same confound ([23] Helms 2014, Level 5). The [24] Burke 2011 carbohydrate-intake framework reached the same conclusion from the carbohydrate side ([24] Burke 2011, Level 5). The [29] Zourdos 2016 RIR-RPE scale in J Strength Cond Res placed the same on the perceived-exertion side: the RIR-RPE scale is most reliable when one variable is changed at a time; confounded sessions produce ambiguous RPE-reads ([29] Zourdos 2016, Level 2b).

The honest read for the rower: the one-variable-at-a-time rule is the diagnostic, not the diagnosis. The [11] Garber 2011 ACSM position stand placed the rule on the canonical-progression side: incremental dose-response is the cornerstone of prescription; the rule is the constraint that makes the dose-response evidence usable ([11] Garber 2011, Level 5). The [16] Issurin 2008 block-periodization review placed the same on the block-organization side: a focused block runs across weeks, and the variable being tested is held through the block; the rule is the constraint that makes the block legible ([16] Issurin 2008, Level 5). The [18] Bompa 2019 periodization textbook reached the same conclusion from the textbook side: the one-variable-at-a-time rule is the operational implementation of the periodization sequence ([18] Bompa 2019, Level 5).

The multi-modal signal in practice: HR, sRPE, duration drift

The multi-modal signal has three channels. The [4] Halson 2014 training-load monitoring review placed the multi-channel signal on the diagnostic side: HR (heart-rate-based fatigue markers), sRPE (session rating of perceived exertion), and duration drift (the rower's actual duration vs. the prescribed duration) are the three channels, and the AI coach reads all three ([4] Halson 2014, Level 5). The [10] Buchheit 2014 HR-monitoring review placed the same on the HR-side: HR-derived fatigue markers include resting HR, HR variability, and HR-recovery; each is a different read on the same underlying fatigue ([10] Buchheit 2014, Level 5). The [1] Foster 2001 session-RPE method placed the sRPE channel on the load-monitoring side: sRPE × duration is the load calculation, and sRPE is the rower's self-report ([1] Foster 2001, Level 5). The [28] Murtagh 2018 rowing-specific load-management review placed the duration-drift channel on the rowing-specific side: when prescribed duration is 30 minutes and the rower rows 28, the duration-drift channel catches the regression ([28] Murtagh 2018, Level 1a).

The [7] Plews 2018 evaluating-adaptation paper placed the multi-modal signal on the HRV side: HRV is the primary channel, and sRPE + duration drift are the secondary channels; the rower who reports all three gives the AI coach the diagnostic ([7] Plews 2018, Level 1b). The [8] Vesterinen 2016 HRV-guided field trial placed the same on the field-trial side: HRV-guided prescription reads the rolling HRV trend and adjusts the next session; the rower who reports HRV daily gives the AI coach the adaptive-prescription signal ([8] Vesterinen 2016, Level 1b/2b). The [11] Garber 2011 ACSM position stand placed the same on the canonical-progression side: dose-response curves for cardiorespiratory fitness are validated by multi-modal signals, and the rower who holds one variable at a time produces a clean multi-modal signal ([11] Garber 2011, Level 5).

The multi-modal signal is not a single number. The [5] Borg 1982 CR-10 RPE scale placed the RPE-channel on the categorical-perception side: CR-10 is a categorical scale, and the rower's CR-10 read is most reliable when one dial is being changed ([5] Borg 1982, Level 5). The [29] Zourdos 2016 RIR-RPE scale placed the same on the perceived-exertion side: RIR-RPE is most reliable when one dial is being changed, and confounded sessions produce ambiguous RIR-RPE reads ([29] Zourdos 2016, Level 2b). The honest read: each channel is a partial read; the constellation is the diagnostic.

The [6] Impellizzeri 2019 injury-and-illness-prevention review placed the multi-modal signal on the chronic-load side: chronic-vs-acute load is the safety channel, and the multi-modal signal includes chronic-vs-acute as a fourth input ([6] Impellizzeri 2019, Level 1a). The [22] Sawka 2007 ACSM exertional-heat-illness position stand reached the same conclusion from the safety side: the multi-modal signal includes a heat-stress channel during hot sessions, and the AI coach reads heat-stress as a fourth input ([22] Sawka 2007, Level 5). The [2] Banister 1975 TRIMP framework placed the same on the load-product side: TRIMP = duration × intensity, and the multi-modal signal collapses to TRIMP plus the chronic-load adjustment ([2] Banister 1975, Level 5).

Common mistakes: the six ways the rule gets violated

The rule gets violated in six common ways. The first is adding two dials in the same session because the rower feels fresh and wants to push. The [15] Kiely 2012 periodization critique placed this on the evidence side: a confounded session cannot tell which dial produced the result, and the rower cannot tell whether the next session will hold ([15] Kiely 2012, Level 5). The [4] Halson 2014 training-load monitoring review placed the same on the multi-modal-signal side: the signal is uninterpretable when two inputs are changing ([4] Halson 2014, Level 5).

The second is changing the watch-period window because the rower is impatient. The [8] Vesterinen 2016 HRV-guided field trial placed the multi-week window on the adaptive-prescription side: HRV-guided adjustment works on a multi-week horizon because adaptation has its own timeline ([8] Vesterinen 2016, Level 1b/2b). The [9] Mann 2014 high-vs-low-responder study placed the same on the individual-variation side: high-responders adapt faster than low-responders; cutting the watch period because one rower adapts fast is a misread of the multi-responder average ([9] Mann 2014, Level 2b). The honest read: the watch period is the rule; impatience is not an exception.

The third is treating pace as a substitute for duration because the rower wants a faster result. The [25] Swain 2006 vigorous-vs-moderate review placed pace on the dose-response side: pace is the most intensity-sensitive dial, and the dose-response curve for pace is steeper than the dose-response curve for duration ([25] Swain 2006, Level 5). The [3] Seiler 2010 polarised-training framework placed the same on the endurance side: low-intensity duration carries most of the endurance-training load; pace cannot substitute for duration ([3] Seiler 2010, Level 1a/2a). The honest read: pace is the last dial for a reason; substituting pace for duration trades adaptation for fatigue.

The fourth is reading the single-session result because the rower had a great session. The [10] Buchheit 2014 HR-monitoring review placed the single-session read on the trend-monitoring side: HR-derived fatigue markers fluctuate day-to-day; the single-session read is noisy ([10] Buchheit 2014, Level 5). The [7] Plews 2018 evaluating-adaptation paper placed the same on the HRV side: HRV-based fatigue markers take days to settle, and the variable-change effect takes longer to manifest ([7] Plews 2018, Level 1b). The honest read: a great session is a single data point; the trend is the read.

The fifth is resetting the dial when the trend is positive because the rower thinks more change is faster. The [11] Garber 2011 ACSM position stand placed the dose-response curve on the canonical-progression side: a positive trend means the dial is producing adaptation; resetting the dial throws away the adaptation ([11] Garber 2011, Level 5). The [16] Issurin 2008 block-periodization review placed the same on the block-organization side: within a focused block, the dial is held through the block ([16] Issurin 2008, Level 5). The honest read: a positive trend is a reason to keep the dial, not to change it.

The sixth is abandoning the dial when the trend is flat because the rower assumes the dial is wrong. The [9] Mann 2014 high-vs-low-responder study placed the flat-trend case on the individual-variation side: rower-to-rower response varies; a flat trend at week 2 may turn positive at week 3 ([9] Mann 2014, Level 2b). The [6] Impellizzeri 2019 injury-and-illness-prevention review placed the same on the chronic-load side: chronic load takes weeks to settle; the acute window can be misleading ([6] Impellizzeri 2019, Level 1a). The honest read: a flat trend at week 2 is a reason to wait, not a reason to abandon.

The summary in one paragraph

Progress in indoor rowing comes from changing one variable at a time. The four dials are duration, repetitions, recovery, and pace; pace is the last to touch. The [1] Foster 2001 session-RPE method placed the rule on the operational side ([1] Foster 2001, Level 5). The [2] Banister 1975 TRIMP framework placed the four dials on the load side ([2] Banister 1975, Level 5). The [3] Seiler 2010 polarised-training framework placed duration on the largest-load side ([3] Seiler 2010, Level 1a/2a). The [4] Halson 2014 training-load monitoring review placed the multi-modal signal on the diagnostic side ([4] Halson 2014, Level 5). The [5] Borg 1982 CR-10 RPE scale placed perceived exertion on the second-channel side ([5] Borg 1982, Level 5). The [6] Impellizzeri 2019 injury-and-illness-prevention review placed the chronic-vs-acute-load calculation on the safety side ([6] Impellizzeri 2019, Level 1a). The [7] Plews 2018 evaluating-adaptation paper placed HRV on the diagnostic side ([7] Plews 2018, Level 1b). The [8] Vesterinen 2016 HRV-guided field trial placed adaptive prescription on the field-trial side ([8] Vesterinen 2016, Level 1b/2b). The [9] Mann 2014 high-vs-low-responder study placed rower-to-rower variation on the individual side ([9] Mann 2014, Level 2b). The [10] Buchheit 2014 HR-monitoring review placed the trend-monitoring side on the diagnostic side ([10] Buchheit 2014, Level 5). The [11] Garber 2011 ACSM position stand placed incremental dose-response on the canonical-progression side ([11] Garber 2011, Level 5). The [12] Pescatello 2021 ACSM Guidelines placed clinical prescription on the dose-response side ([12] Pescatello 2021, Level 5). The [13] Tanaka 2001 HRmax formula placed pace on the rate-cap side ([13] Tanaka 2001, Level 1b). The [14] Karvonen 1957 HR-reserve formula placed HR reserve on the rate-band side ([14] Karvonen 1957, Level 5). The [15] Kiely 2012 periodization critique placed the evidence side on the dose-response side ([15] Kiely 2012, Level 5). The [16] Issurin 2008 block-periodization review placed block organization on the focused-block side ([16] Issurin 2008, Level 5). The [17] Matveyev 1981 fundamentals of sports training placed multi-year development on the long-horizon side ([17] Matveyev 1981, Level 5). The [18] Bompa 2019 periodization textbook placed the periodization sequence on the textbook side ([18] Bompa 2019, Level 5). The [19] Kraemer 2004 progression-prescription framework placed the prescription sequence on the canonical side ([19] Kraemer 2004, Level 5). The [20] Rietjens 2001 rowing-specific periodization paper placed rowing progression on the rowing side ([20] Rietjens 2001, Level 1b/2b). The [21] Platonov 2006 multi-year periodization framework placed long-horizon programming on the multi-year side ([21] Platonov 2006, Level 5). The [22] Sawka 2007 ACSM exertional-heat-illness position stand placed safety on the heat-illness side ([22] Sawka 2007, Level 5). The [23] Helms 2014 nutrition-and-training framework placed nutrition progression on the parallel side ([23] Helms 2014, Level 5). The [24] Burke 2011 carbohydrate-intake framework placed carbohydrate adaptation on the metabolic side ([24] Burke 2011, Level 5). The [25] Swain 2006 vigorous-vs-moderate review placed pace on the dose-response side ([25] Swain 2006, Level 5). The [26] Pollock 1978 dose-response review placed dose-response on the foundational side ([26] Pollock 1978, Level 5). The [27] Astrand 2003 textbook of work physiology placed work physiology on the foundational side ([27] Astrand 2003, Level 5). The [28] Murtagh 2018 rowing-specific load-management review placed rowing progression on the rowing-specific side ([28] Murtagh 2018, Level 1a). The [29] Zourdos 2016 RIR-RPE scale placed perceived exertion on the RIR-RPE side ([29] Zourdos 2016, Level 2b). The [30] ACSM 2009 progression-models position stand placed incremental dose-response on the canonical-progression side ([30] ACSM 2009, Level 5).

The right posture is to pick one dial, hold the others constant, make the change, and watch the trend for at least two weeks. The AI coach chooses the dial and holds the others. Duration is the default; repetitions are next; recovery is third; pace is last. The one-variable-at-a-time rule is the operational implementation of the periodization sequence, and the two-week watch period is the discipline that turns the rule into measurable progress.

For a deeper exploration of how the rule interacts with the rate-cap diagnostic, see our rate-caps guide and our rate-ladders diagnostic.

What to do with this article

Read the principle: progress comes from changing one variable at a time. The [1] Foster 2001 session-RPE method places this on the operational side; the [15] Kiely 2012 critique places it on the evidence side; the [11] Garber 2011 ACSM position stand places it on the canonical-progression side.

Read the four dials: duration, repetitions, recovery, pace. The [26] Pollock 1978 dose-response review places duration on the foundational side; the [25] Swain 2006 review places pace on the dose-response side; the [3] Seiler 2010 framework places duration on the largest-load side.

Read the rule: pace is the last dial. The [22] Sawka 2007 ACSM heat-illness position stand places this on the safety side; the [28] Murtagh 2018 rowing-specific review places it on the rowing-specific side; the [3] Seiler 2010 polarised-training framework places it on the endurance side.

Read the watch period: two weeks minimum. The [8] Vesterinen 2016 HRV-guided field trial places this on the adaptive-prescription side; the [6] Impellizzeri 2019 review places it on the chronic-load side; the [11] Garber 2011 ACSM position stand places it on the canonical-progression side.

Read the practical read: pick the dial, hold the others, make the change, watch the trend for two weeks, read the multi-modal signal. The [4] Halson 2014 training-load monitoring review places this on the multi-modal-signal side; the [16] Issurin 2008 block-periodization review places it on the block-organization side.

When the trend settles, the AI coach moves to the next dial. The one-variable-at-a-time rule is the operational implementation of the periodization sequence; the watch period is the discipline that turns the rule into measurable progress.