Engineered muscle force: a normalised reference table

A force specification written as a bare number in micronewtons is not a specification. Engineered muscle force scales with cross-sectional area, and constructs in the published literature differ in cross-section by more than four orders of magnitude — from about 4.5 × 10⁻⁵ mm² for a myotube on a cantilever to about 5.7 mm² for a rat myobundle. Two constructs producing the same absolute force can differ a thousand-fold in the quality of the muscle.

The comparable quantity is specific force: contractile force divided by cross-sectional area, expressed in mN/mm². That unit is numerically identical to kPa and to kN/m², all three of which appear in the literature. This page normalises published values into that unit, keeps the absolute force alongside it, and states the measurement platform and stimulation condition for every row.

The reference points to hold in your head

Engineered muscle is therefore, at its published best, roughly one tenth of native. Most of it is one fiftieth to one hundredth. Any specification that implicitly assumes native performance is outside the demonstrated envelope.

The normalised table

Rows marked [V] are drawn from the systematic normalisation published by Vesga-Castro and colleagues in eLife (2022), which recalculated specific force from reported dimensions across the literature; in that table several values were themselves recalculated or extracted from figures rather than read from text, and those caveats propagate here. The author-and-year labels on those rows are as given in that review, and we have not independently confirmed each primary paper’s DOI — the checkable citation for a [V] row is the review. Rows marked [P] were read directly from the primary paper by us, and those papers are listed individually in the sources below. Specific force is stated in mN/mm², equivalent to kPa.

Force transducer measurements — 3D constructs

Post / pillar deflection measurements

Cantilever / thin-film measurements — 2D and micro-scale

Biohybrid robot actuators specifically

How to read this table without misleading yourself

Cross-section is measured inconsistently, and it changes the answer by more than the cell source does. The clearest illustration is Juhas and Bursac 2014: the same rat myobundle gives 11.89 mN/mm² tetanic when normalised to total bundle cross-section and 43.39 ± 3.82 mN/mm² when normalised to the F-actin-positive fraction — a 3.6-fold difference from a definitional choice. Hinds et al. 2011 shows the same effect: 0.50 mN/mm² by bundle cross-section against 9.4 ± 0.7 mN/mm² effective. When you compare two published specific forces, first check whether both used total cross-section.

Measurement platform sets the accessible range. Vesga-Castro and colleagues give the reported working ranges: roughly 0.02–3.31 µN for cantilevers, 1.4 µN to 1.1 mN for micro-posts, and 18 µN to 30 mN for force transducers. A cantilever cannot measure a myobundle and a transducer cannot resolve a single myotube. Comparing across platforms compares partly the platforms.

Absolute force without cross-section is not normalisable. The Smith 2022 rows illustrate this honestly — they are among the most useful absolute measurements in the field, on a commercial platform, but the paper does not report tissue cross-sectional area alongside the force values, so no specific force can be derived and we have not invented one.

Anchor stiffness is part of the measurement. In the same study, day-10 twitch force fell from 1156.3 ± 129.1 µN at 1.92 N/m post stiffness to 624.0 ± 154.6 µN at 0.96 N/m, and at 0.16 and 0.48 N/m the posts bent so far under passive tension that active contraction could not be resolved at all. A force number without an anchor stiffness is incomplete.

Twitch and tetanic are different measurements. Tetanic-to-twitch ratios in the normalised literature run roughly from 1 to over 5, and the ratio itself carries information about maturation. Specifying “500 µN” without saying whether that is twitch or tetanic leaves a factor of two to five undefined.

What moves specific force

Ranked by the size of the documented effect:

Chronic electrical stimulation. The largest single lever. One week of intermittent 1 Hz stimulation increased tetanic force approximately three-fold in human myobundles and roughly doubled specific force, from 9.1 ± 0.38 to 19.3 ± 0.63 mN/mm²; a 10 Hz protocol gave 18.9 mN/mm², statistically comparable. The stimulated result is the highest specific force reported for engineered human muscle. The same paper notes that stimulated myobundles showed decreased fatigue resistance, so the improvement is not free.

Mechanical loading during maturation. Chen and colleagues’ self-training approach — muscle pairs loaded against each other during differentiation — produced 7.05 mN tetanic against 3.70 mN for a high-stiffness control and 0.86 mN for the low-stiffness configuration they describe as most commonly used in biohybrid robotics. That is an eight-fold spread within one cell source from maturation mechanics alone.

Time in culture, up to a plateau. iPSC-derived engineered muscle tissues increased from 385.8 ± 51.1 µN twitch at day 7 to 1156.3 ± 129.1 µN at day 10, then plateaued at 943.1 ± 162.1 µN by day 14 with no further improvement. Primary human myoblast tissues in the same study were still rising at day 14. Culturing longer is not reliably better and can be worse.

Media composition. In the same work, 2% horse serum throughout produced greater force development than reverting to 15% fetal bovine serum after day 7, or than a serum-free knockout serum replacement formulation.

Cell source. Real but secondary. The highest specific forces in the normalised table come from hPSC-derived human myoblasts (Xu et al. 2019, 48.73 mN/mm² tetanic on a 0.06 mm² cross-section) and from rat myobundles normalised to the myotube-positive fraction — but the C2C12 range alone spans 0.06 to 8.51 mN/mm², which overlaps almost everything else.

Writing a force specification that is answerable

A specification that a supplier or collaborator can actually quote against needs all seven of these:

1. Twitch or tetanic, stated explicitly.
2. Absolute force, with units.
3. Cross-sectional area, either as a requirement or as the basis on which force will be normalised — and which definition of cross-section applies.
4. Anchor stiffness, or the geometry that determines it.
5. Stimulation protocol at which the force must be demonstrated: waveform, amplitude, frequency, pulse width.
6. Timepoint at which the measurement is taken, and whether the construct must still meet it later.
7. Measurement method, since cantilever, post deflection and transducer results are not interchangeable.

A specification containing all seven can be checked against this table before anyone spends money. One containing only item 2 cannot.

Is a given specification achievable

Work it through against the table. A requirement of 500 µN twitch, for instance, sits inside the demonstrated envelope on several routes: iPSC-derived engineered muscle tissue on 1.92 N/m posts exceeded it at day 7 (385.8 µN) and doubled it by day 10 (1156.3 µN); primary human myoblast tissues reached 568.0 ± 103.7 µN twitch at day 14 on the same platform. A requirement in the low mN range is achievable but sits at the top of the published C2C12 literature and generally requires a purpose-built maturation protocol. A requirement above roughly 20 mN/mm² specific force in human tissue has been reported only twice in the normalised table and should be treated as a research programme, not a purchase.

And a requirement anywhere near native human muscle — 150–250 mN/mm² — is not currently achievable by anyone.

No supplier sells against these numbers

Worth stating plainly, because this table reads like a product specification and is not one. No vendor surveyed lists a finished, force-specified engineered muscle construct as a catalogue item. The organisations that build such constructs — Curi Bio, eNUVIO, Creative Biolabs — deliver reports and datasets. Whether any of them will contract to a force acceptance criterion on a physical part is a commercial question. The CRO directory covers who to ask, and the sourcing guide covers what the inputs cost.

Corrections welcome

This table is intended to be the reference version, which means it needs to be right. If a row misstates a value, uses the wrong cross-section convention, or omits a study that belongs here, tell us and we will correct it with the citation attached. Values recalculated by the source review rather than read from the primary text are marked as such deliberately, and we would rather replace them with directly read values where those exist.