Which Finger-Drumming Tempo Is Repeatable Today? A Reversible Tempo-Ladder Test

“Repeatable today” needs an accuracy condition

A tempo number describes the requested rate, not whether the intended pattern survived. Pfordresher, Palmer, and Jungers tested experienced pianists playing unfamiliar sequences at eight tempi and found a speed–accuracy trade-off when performers approached their limits. That result does not give finger drummers a threshold, but it does justify recording errors alongside tempo instead of treating the highest completed take as the answer.

The contract can be strict without pretending to scientific precision. It should answer: Were all intended hits present and in order? Were there extra or wrong-pad hits? Did the loop return to beat 1? Did a written accent or hand assignment survive? If timing is part of the target, use the resolution your recorder honestly exposes and state it. Do not infer a millisecond score from a coarse visual grid.

Freeze one pattern before testing tempo

Use a pattern that is already understood at a comfortable rate. This test is not the place to learn new notes, change pad layout, and build speed simultaneously. Caramiaux and colleagues found that tempo variability and its schedule could affect motor learning and transfer differently from timing skill in novice piano practice. Their protocol does not prescribe this ladder; it is a reason to declare the tempo order and avoid silently mixing multiple learning variables.

1. Write one bar: mark every onset, rest, simultaneous pair, accent, and loop boundary. If the pattern has an optional fill, remove it for this test.
2. Freeze movement: declare the pad layout and finger assignment. A deliberate substitution can be a later experiment, but an unrecorded substitution changes today’s task.
3. Freeze sound and route: keep the kit, pad sensitivity or fixed-velocity state, monitoring path, click subdivision, and recording route unchanged.
4. Disable helpers that perform the test for you: turn off note repeat, arpeggiation, input quantization, groove, swing, and timing humanization. Preserve raw MIDI or an unedited audio/video take.
5. Choose the pass length: play four uninterrupted loops of the same one-bar pattern, then record only the next bar’s beat-1 anchor. The anchor makes the final boundary audible; it is not a fifth loop.
6. Choose one step: use the smallest tempo adjustment your tool can reproduce and that is practical for the session. Record it as Δ; do not change it midway.
7. Cap the ordered test: declare one fixed rest between valid passes and a maximum number of ladder rungs before starting. Reaching the cap ends the climb even if every rung passes.

Build a five-code error ledger

Listen and inspect in the same order after every pass. Mark the first failed gate, but keep the raw take so a later review can disagree. A take either satisfies today’s written contract or it does not; do not relax the contract only because the tempo number is attractive.

If a dropout, duplicate trigger, changed route, notification, or accidental tempo edit invalidates the take, mark X for setup failure and repeat the same stage. An X take stays in the ledger but does not count toward the two valid passes. A second X at the same stage aborts the remaining sequence as setup-unresolved. If it happens before the terminal rung, do not define R; wherever it happens, do not record later planned stages or schedule a follow-up target. Repair the setup and begin a fresh session. If discomfort appears, stop the session; do not convert it into another error code.

B0 proves the task exists before the climb

Choose a starting tempo that feels comfortably controllable. “Comfortable” is personal and pattern-specific, so this article supplies no number. Record B0a and B0b, each with four complete loops plus the beat-1 anchor and the declared rest between passes. Both must satisfy the same contract before the ladder begins.

Key: B = baseline/return, L = ladder rung, R = terminal rung, F = same-gate failure twice, U = unresolved split or mixed result, P = passed predeclared cap, C = one-rung-back control, and Δ = your declared increment. These labels define an observation order, not a prescribed training dose.

If B0 fails, do not lower the tempo repeatedly until the page can be completed. First ask whether the pattern is actually learned, the layout is stable, and the contract is realistic. Record the lower discovery tempo as a new baseline only after stating the change; otherwise later sessions will compare different tasks under one label.

Climb only after two valid passes; stop at the declared terminal rung

1. Set L1 to B0 + Δ: record two new valid passes, each with four loops plus the anchor. Do not reuse or time-stretch B0.
2. Advance only if both pass: a pass followed by a fail—or a fail followed by a pass—makes this terminal rung U for unresolved. Do not retry selectively; end the climb.
3. Continue in the same order: add exactly Δ for L2, L3, and later rungs. Keep the declared rest, pass length, click subdivision, and all other settings unchanged.
4. Stop at F: when the same gate fails on two consecutive passes, call it today’s first repeatable-failure rung. Do not take “one more shot” to rescue the number.
5. Keep mixed failures honest: if both passes fail but the first code differs—such as E then LB—label the terminal rung U, not F. Do not invent a single diagnosis.
6. Honor the cap: if both passes at the predeclared final rung pass, label that terminal rung P and stop. Report “tested through P; no failed rung located,” not a maximum.

BOSS’s DB-90 offers product-specific Gradual Up/Down and Step Up/Down modes that increase and then decrease tempo; its documented increments and durations belong to that device. PAS likewise recommends practicing rudiments from open (slow) to close (fast) to open. Those sources support making the return visible, but neither supplies a universal finger-drumming increment or this article’s pass gates. Because every rung adds exposure and the order is always upward, this protocol cannot separate warming up, short-term practice, fatigue, and order; its result applies only to the declared run.

Back off from R, then return to B0

After terminal rung R—whether F, U, or passed cap P—reduce the tempo by exactly Δ and record C1 and C2. For F or U, R−Δ is the preceding rung that had passed twice; for P, it is simply the one-rung-back control. If C fails, do not keep descending and claim a precise boundary; attention, setup, movement, or task memory may have changed. Save the ledger as unresolved.

Finally return to the original B0 and record B1a and B1b. This is the control that a one-way speed run omits:

• If R is F or U and C and B1 both pass, report R−Δ as stable today under this order; report F as repeated same-gate failure or U as unresolved.
• If R is passed cap P and C and B1 both pass, report tested through P under this order; no failed rung located.
• If C passes but B1 fails, the original task did not return. Keep the takes; do not report a new maximum.
• If C fails but B1 passes, the tempo conclusion remains unresolved: baseline control returned, but the one-rung backoff did not.
• If both fail, stop. The session state no longer supports a narrow speed conclusion.

This return does not prove why a failure occurred. It only checks whether performance at known settings can be recovered after the climb. Technique, memory, listening, pad response, and attention remain possible contributors unless tested separately.

A baseline-first no-ramp follow-up tests later reproduction

At the next practice session on another day, restore the same pattern card and route. Record two valid B0 passes first; both must meet the contract to pass this baseline screen, or the follow-up stops. This screen supports continuing the check but does not make the sessions fully comparable, and the higher-rung attempt is not a cold start. Then jump directly—without climbing—to the last rung that passed twice in the prior session and record two valid passes. Keep the previous Δ, rest, pass length, and contract.

The 2018 tempo-variability study separated motor and timing outcomes and found no single practice schedule that dominated every measure. This baseline-first check is therefore only a later reproduction test for this exact task, not proof of retention or a promise that one ladder schedule generalizes to every groove, layout, or player.

Tempo-ladder questions

How many BPM should I add per rung?

There is no universal increment here. Choose one reproducible adjustment small enough to locate a change without creating an impractically long session, write it down, and keep it fixed. A hardware coach’s factory increment is a product behavior, not an evidence-based prescription for every player.

Should I keep trying a failed rung until I get one clean take?

No. That changes the exposure and lets selection bias choose the most flattering attempt. Under this framework, the climb stops whenever both valid passes do not pass: the same-gate failure twice is F, while a split or mixed result is U.

Can I quantize the MIDI before checking it?

Not for the practice decision. Quantization can move the evidence used to judge timing and boundaries. Preserve the raw take; if production needs editing, quantize a named copy afterward.

Is playing double-time the same as increasing BPM?

No. Double-time can change the perceived subdivision or performance layer while the underlying tempo stays fixed. This test changes the declared metronome tempo for one unchanged pattern. Use the separate half-time / double-time guide to practice feel without confusing it with this ladder.

Do I need MIDI recording?

No. Raw MIDI makes event order and onset inspection convenient, but unedited audio or video plus a written ledger can support the gates. Use only the precision the recording actually reveals.