What we measured. And what it is worth.
This document does not set out to convince. It sets out to be verifiable. What we measured in the field, with which instruments, under which conditions — and above all what those conditions do not allow us to conclude.
The limits of our observations take up as much room here as the results. That is deliberate. The controlled protocols we plan are described in this document, with their state of progress — we will publish the results, including negative ones.
6 subjects, 4 muscle groups. It is preferentially the weaker side that improves.
Individual gain percentages are not interpretable until the detection threshold of the protocol is established. They are therefore not used as headline figures: they are recorded further down, in the detailed measurements, with their status.
Three levels of evidence — and nothing above them
Every observation on this page carries its level of certainty. An exploratory result never moves up a rung.
Physics or published literature
Verifiable by a third party, independently of us. It is the only level we rely on without reservation.
Measured by us, with instruments
Under uncontrolled conditions. Descriptive — not conclusive. Almost all of this document sits here.
A coherent proposition, under investigation
Stated because it guides the research. Not demonstrated, and presented as such.
What “uncontrolled” means — precisely
The “without” measurement always precedes the “with” measurement. A simple learning or warm-up effect may therefore contribute to the result.
No indistinguishable placebo condition was used in the strength series.
The subject and the operator both know what is being worn. Expectation can influence a voluntary performance.
We have not yet measured the smallest difference our protocol allows us to claim as real. That is a gap, and we write it rather than let it be guessed. The reliability study that will produce this figure is the first protocol we will run.
A parameter we did not invent.
Before saying what we measure, we have to say why this parameter. Left/right strength asymmetry is an established clinical object, with its thresholds, its tests and its decisions.
In return to sport
Passing the full return-to-sport battery — score > 90 on the KOS-ADL, the global rating of function, quadriceps strength symmetry and hop tests — was associated with 5.6% reinjury versus 38.2% on failure (HR 0.16), without reaching statistical significance (p = 0.075). More symmetrical quadriceps strength before return did significantly reduce the reinjury rate.
Failing to meet the discharge criteria, symmetry among them, is associated with a fourfold risk of graft rupture.
In low back pain
The most important body of workIn acute unilateral low back pain: multifidus asymmetry of 31% ± 8% on the symptomatic side, against less than 5% in healthy subjects. The atrophy matched the symptomatic vertebral level in 24 patients out of 26.
Multifidus recovery is not automatic after resolution of a first episode of acute low back pain.
Multifidus volume reduced by 18.1% at the L5-S1 level, in active individuals.
The pain goes away.
The asymmetry stays.
It has been published since 1996, and it is the reason we measure this parameter rather than another: it persists after the symptom has gone, and it can be tracked over time.
Schematic, after the pattern described by Hides et al. (1996). The curves illustrate the dissociation between the two parameters; this is not a data plot, and the values are not to scale.
The limits
Including the work that contradicts usHides 1994 states it directly: no correlation between the degree of asymmetry and the severity of symptoms. More asymmetry does not mean more pain.
The direction of causality is not established. Does asymmetry promote pain, or does pain inhibit the muscle? Reflex muscle inhibition linked to pain is well documented, and argues rather for the second direction.
The 10-15% threshold used in sport is not always substantiated: fifteen articles out of eighteen use it without correctly referencing its origin, and the reported effects on injury and performance are inconsistent.
Findings were highly inconsistent; the authors conclude that no clear recommendation can be made on the relationship between asymmetry and injury. Some included cohorts report an association around 15%, others do not.
No increase in hamstring injury risk at 10, 15 or 20% asymmetry — a result contradicting other work.
They forbid claiming that asymmetry causes pain. We do not claim it.
Asymmetry remains measurable, it persists, and restoring it is a recognised rehabilitation objective.
It is not the magnitude. It is the pattern.
A series of isometric strength tests on a Kinvent dynamometer, across 6 independent subjects and 4 muscle groups.
And above all: it is preferentially the weaker side that improves.
An ordinary warm-up lifts both sides comparably. That is not what we observe here: it is the weaker side that improves the most, systematically, across 6 subjects and 4 different muscle groups. The result is not an overall increase — it is a rebalancing.
Shoulder abduction — Jon
Improvement on both sides, more marked on the left.
Knee extension — Stéphane
Weaker side: strongest improvement.
Knee flexion — Élodie
Weaker side: strongest improvement.
The same pattern across the other subjects
A learning effect can also be asymmetric — a side less used day to day may have more room to improve on a second attempt. We cannot yet rule that out.
The individual magnitudes observed are large, sometimes very large. Until the detection threshold of our protocol is established, we do not present them as results: we present the pattern.
The answer to both objections is the same — a randomised, blinded replication against a sham circuit. It is planned.
Regression to the mean
Measuring a left/right gap twice in a row statistically produces a decrease, with no intervention at all.
This explanation accounts both for the reduction in asymmetry and for the fact that the weaker side improves most — that is, exactly the pattern we observe. We cannot rule it out as things stand.
It can only be ruled out by three design corrections
Randomise the order of conditions. No condition may be systematically measured second.
PlannedRun blinded, against a sham circuit. Neither subject nor operator may know what is worn.
PlannedDetermine the weaker side in an earlier session. The side of interest must not be picked by the same day's measurement.
PlannedPosture, range of motion, maximal strength, physiology
Each with its instrument, its level of evidence and its methodological limit.
Postural sway in the squat
Sway area at equal load, on a K-Force Plates force platform.
Cervical rotation range
Measured with K-Move, before / after calibration.
Maximal strength — ARX machine
In professional athletes, recorded immediately after calibration and compared with the best personal score on record for the machine. R15 Studio, Paris.
Methodological limit. A personal best is the best of a large number of attempts. Comparing it with a single measurement is not a balanced comparison. This observation will remain indicative until it has been repeated in a protocol comparing two series of attempts under identical conditions.
Physiology and cortical activity
EEG, heart rate, training load, recovery. Single subject — shown to illustrate the instruments, not to conclude.
The cervical test included an additional condition: inactive circuits. This observation points in the opposite direction to what a simple expectation effect would predict.
It does, however, cover a single test, without randomisation or blinding, and an inactive circuit is not a validated sham: we cannot guarantee it was indistinguishable to the subject. We mention it because it guides the research, not because it concludes anything.
A personal record is the best of a great many attempts. Comparing it with a single measurement is not a balanced comparison: it is not the same kind of value.
This observation will remain indicative until it is repeated in a protocol comparing two sets of attempts under identical conditions.
Single-subject observations — exploratory
Four biophysical profiles — not a flaw, but the logic
The magnitude of the response varies considerably from one person to another. Our field observations led us to identify four distinct response configurations.
The engraved plates name the register. Omega exists in two orientations — arrow up and arrow down are two distinct calibrations. Go is a separate range, with no profile determination.
A calibration that does not match a person’s profile yields results below what they could obtain. That is why the profile is determined beforehand, and is not chosen.
The Perfect Posture Test — 15 minutes, with a certified tester — identifies the profile and enables the matching calibration.
A figure that changes the perspective
Profile distribution observed in two very different settings.
A system in which choosing the wrong profile explains a failure is hard to catch out. And a fourth profile that takes in precisely the cases the other three do not explain makes that objection stronger still. We name it for that reason.
Our chrysalis-state hypothesis does, however, carry a consequence that can be tested. If Harmonia is a stage and not a category, some of the people it covers must show a defined profile on a later test. A retest at six and twelve months answers the question. If the conversion rate is zero, the hypothesis is wrong, and we will publish that.
Across roughly 5,000 sessions since 2023, we record on the order of one session in two hundred with no correction across the three planes under our current criteria — criteria that were not pre-registered. Our answer remains a pre-registered study, whose plan must explicitly include the Harmonia case. Without that, the fourth profile would become an adjustment variable absorbing every failure.
We predict that the response to the circuit matching the profile will be greater.
If it is not, our profile grid is wrong — and we will publish that.
The most important part of this document
A trained reader looks first for what is missing. Here is our list, before anyone asks for it.
About one person in two hundred does not fall under any of the three calibrated profiles.
When the Perfect Posture Test identifies one of the three — Q-Alpha, Q-Theta or Q-Omega — the postural correction we observe spans all three planes. In roughly one person in two hundred, none of the three produces that result. Those people fall under the fourth profile, Harmonia, and receive the universal signature — the one used in Q-Go.
Harmonia is not an exception, it is a state: a system that has not yet differentiated toward one of the three registers. We call it the chrysalis state, and the image says what it should — a stage, not a defect.
What we do not know: whether that state evolves. If the hypothesis holds, some of these people should show a defined profile on a later test. We have not measured this. It is one of the reasons profile stability over time sits among our open questions.
We say it before the sale, not after.
Outside a controlled protocol, without systematic counting. We will publish this rate as soon as a standardised series allows it to be established.
It is also what makes the universal signature recommendable without a test. It is optimal for no one and compatible with everyone: it is the only one of the four whose assignment does not depend on a prior determination.
What is the smallest difference our protocol can detect?
A test-retest reliability study, within- and between-operator, is the first protocol we will run. It has not started. Until that figure exists, we cannot claim that any given difference is significant.
Does the effect survive blinding?
Sham-controlled, double-blind protocols are in preparation, with an independent third party holding the key alone.
Is a person’s profile stable over time?
A retest at six and twelve months would answer the question. The protocol is described; it has not started.
By what mechanism?
And not only for us: the debate on the action of weak fields on living systems is open in the scientific literature. We therefore base nothing on it.
The research programme
Our three methodological commitments
Protocols are written before the first measurement — question, single primary metric, sample size, analysis plan.
An independent third party prepares the coded conditions and holds the key alone. Without that, a study is not blinded.
We will publish the results, including negative ones. A null result is information.
Public, verifiable — and not ours
The principle our circuits rest on is not a secret. It is in the public domain.
Any laboratory can build the example circuit described in the patent and test the principle. Reproducing Q-Technology's own calibrations, however, requires confidential access to the setting — for instance within an independent study under a non-disclosure agreement. We invite them to.
Every circuit is checked individually before shipping. Our profiles are mutually discriminable by instrument: a circuit of one profile does not respond to the control device set to another profile. Calibration therefore produces physically distinct objects.
