## Main

Smart textiles, capable of sensing and responding to stimuli from the environment1,2, are gaining increasing attention across fields as diverse as health care2,3, sports4 and fashion5, motivated by opportunities in, for example, wearable technology6,7,8,9 and robotics6,10, biosensing11,12, data collection13,14 and information processing11,12,15. The actual textile often remains passive, acting as a carrier for electronics that provide functionality but also require a complex device architecture and a power supply, sometimes inhibiting wearing comfort and washability. If the textile fibres themselves are made from a durable responsive material, doors open for fully autonomous smart textiles.

The advancing class of mechanochromic polymers16, which change their visual appearance in response to mechanical deformation, can be very powerful in this context if the right type is used. Molecular-scale processes may give too low contrast (Raisch et al. recently reported a wavelength shift of Δλ = 12 nm (ref. 17)), and among structurally coloured materials18,19, many rely on the incorporation of nanoparticles. It is unclear if such composites resist regular washing procedures, and the nanoparticle incompressibility limits response range and stretchability, which may not reach the requirements for comfortable wearable devices20. Moreover, the colour response may be less clear-cut, since the cubic symmetry typical of colloidal crystals yields selective reflection along three orthogonal principal directions: a positive uniaxial strain along one direction produces a positive mechanochromic shift Δλ in this direction but a negative shift in the perpendicular plane. Using a crystal of silica nanoparticles, Cheng et al. obtained Δλ ≈ 50 nm, from weakly red to green, before the material broke at 60% strain21. Kim et al. made particle-loaded fibres that elongated 200% before breaking, but the red to green colour shift (Δλ ≈ 0.1 μm) stopped at 80% elongation22. Kolle et al. made one-dimensionally (1D) periodic lamellar fibres by rolling an elastomer bilayer around a glass rod23, but second-order reflections led to colour mixing, and mechanical durability and washability were not assessed, nor was scalability. The removal of the sacrificial glass rod without damaging the elastomers is a non-trivial challenge for fibres that should be long enough to make garments.

A shining star among mechanochromics is the class of cholesteric liquid crystal elastomers (CLCEs)24,25,26,27,28, the self-assembled 1D-periodic helical structure of which gives a fully predictable and reversible mechanochromic response that can span the entire visible spectrum, over a broad elastic range. Using two-dimensional (2D) sheets of CLCE, we recently demonstrated a Δλ = 145 nm shift from red to blue upon 120% strain29, and several other groups have reported a similarly impressive CLCE performance30,31,32,33,34,35,36,37,38,39,40. However, there have been no reports so far of mechanochromic CLCE fibres, to the best of our knowledge. This is because CLCE helix formation is slow, often not reaching the required uniform alignment41 before the Plateau–Rayleigh instability breaks up a one-dimensional fibre of the liquid precursor from which CLCEs are made42,43. Non-polymeric cholesterics have been supported on regular fibres44 or encapsulated within rubber sheaths45, but the liquid state of the cholesteric restrains the mechanochromic response and durability.

Here, we realize CLCE fibres from an oligomeric precursor slightly diluted with solvent, balancing the viscoelastic properties to allow continuous filament extraction, yet delay the Plateau–Rayleigh instability until the helix has developed. We demonstrate a continuous and repeatable mechanochromic response up to Δλ = 155 nm at 200% elongation, and we weave the fibres and sew them into regular fabrics to reveal complex strain patterns, with the fabrics surviving not only long-term repeated use but even several rounds of conventional machine washing.

## Oligomer synthesis and CLCE fibre production

The CLCE precursor is based on an acrylate-terminated liquid crystalline oligomer (LCO), synthesized following a thiol–acrylate Michael addition reaction as shown in Fig. 1a (details in Methods). Triethylamine is used as a catalyst to initiate the click reaction between a liquid-crystal-forming diacrylate monomer (RM257, described in the Methods), a polymerizable chiral dopant (LC756, described in the Methods) and a dithiol chain extender (2,2-(ethylenedioxy) diethanethiol, EDDET). To ensure acrylate termination, the reaction is performed with excess (3 mol%) of acrylate monomer. By varying the catalyst concentration, we prepare LCOs of three different lengths (Methods for details). Unless otherwise stated, fibres are made with the shortest oligomer, LCO1, yielding the highest crosslink density since the end groups are reactive. The LCO is diluted with dichloromethane (DCM) at 20% (for LCO1) by mass to form the precursor. The solvent concentration is key to allowing stable filament formation: too little DCM makes the precursor too hard, such that it fractures upon extension into filament shape, while too much DCM renders it too fluid, allowing the Plateau–Rayleigh instability to deform the filament and even break it into droplets before the helix formation is complete (Extended Data Fig. 1).

The central CLCE reflection wavelength λr is defined by the Bragg equation as41

$${\lambda }_{{\mathrm{r}}}=\bar{n}p\cos \theta ={\lambda }_{0}\cos \theta ,$$
(1)

where $$\bar{n}\approx 1.6$$ is the average refractive index, p is the periodicity (pitch) of the helix, θ is the angle of incidence with respect to the helix axis and λ0 is the maximum λr corresponding to θ = 0, for illumination and observation along the helix axis (retroreflection). To tune the mechanochromic response regime, we reduce p further by dissolving an additional small amount of LC756 into the LCO–DCM precursor mixture. Adding 1.5% or 3.15% (by mass) of LC756, we obtain CLCEs with λ0 in the red or green region, respectively, for a relaxed fibre. Photoinitiator is also added at this stage to enable photopolymerization and crosslinking into the CLCE after the helix development is complete.

As shown in Fig. 1b and Supplementary Video 1, it is easy to manually pull long filaments from the precursor. In order to reproducibly make fibres of much greater length and well-controlled dimensions and mechanochromic properties, a simple homemade set-up is designed as schematically shown in Fig. 1c. A seed filament is extracted from a syringe that delivers CLCE precursor at a set feed rate Q, and it is attached to one end of a rotating mandrel coated with polyvinyl alcohol (PVA). By translating the syringe along the mandrel at a speed v adjusted to the rotation speed ω, a continuous filament is wound onto the mandrel (Supplementary Video 2). The filament diameter can be adjusted from micrometres to millimetres by tuning Q, v and ω. The extraction from the syringe into air is expected to give the filament an initially cylindrical shape with circular cross-section (schematic drawings in Fig. 1d,f), but as the precursor wets the mandrel, it deforms within minutes into a hemicylindrical shape.

Although the precursor at rest is isotropic, the extensional flow during extraction aligns the oligomer uniaxially, yielding a paranematic state with director n (direction of long-range orientational order) along the filament (schematically illustrated in a highly idealized fashion in Fig. 1f), as evidenced by the strong birefringence with the uniform optic axis along the filament immediately after extraction (Extended Data Fig. 2, upper row). As the solvent evaporates, the optical characteristics change, and we see the first strong evidence of coloured selective reflection after about 10 hours (lower row). This reveals that the LCO transitions from the flow-aligned paranematic state to the thermodynamically stable cholesteric state, in which n continuously twists into a helical arrangement. The filament is compressed unidirectionally during solvent evaporation as the wetting to the mandrel prevents shrinkage in the plane29. This ‘anisotropic deswelling’46 flattens the filament into a more belt-like shape (Fig. 1h), a process that promotes helix alignment perpendicular to the belt plane47. After relaxation, the filament is photopolymerized into the final CLCE fibre by ultraviolet (UV) irradiation. The ideal-case structure promoted by anisotropic deswelling is illustrated schematically in Fig. 1g. The overall experimental evidence to be presented below suggests that this is a suitable approximate model for analysing the fibre behaviour, but there are non-negligible local variations.

To assess the helix alignment uniformity, we slice a thin fibre embedded in UV-cured glue (Norland Optical Adhesive, NOA) using a microtome, studying it using polarizing optical microscopy (POM; Fig. 1i) as well as scanning electron microscopy (SEM; Fig. 1j,k). The colour of the slice in POM is uniform over large areas; it shows nearly no birefringence effect when the belt plane is parallel or normal to the polarizer, and with a first-order λ plate inserted, the total birefringence is reduced/increased when the belt normal is along/perpendicular to the slow axis of the λ plate (Supplementary Video 3). This behaviour confirms a helix with submicrometre pitch oriented predominantly perpendicular to the belt plane. In SEM, a periodic set of lines largely parallel to the belt plane are visible (Fig. 1k), corroborating the helix orientation. In the Supplementary Discussion, we consider what may be the origin of these lines, and in Supplementary Note 1, we review the effect of increasing the fibre thickness: while the average helix orientation remains perpendicular to the belt plane for thick fibres also, the quality of alignment decreases.

## Selective reflection and mechanochromic response

We also carry out detailed POM investigations of the intact CLCE fibres: an ~250-μm-wide fibre with red λ0 shown in Fig. 2 and one with green λ0 in Extended Data Fig. 3. The fibres are uniform in width and thickness and show intense selective reflection, clearly visible even with a bright background (Fig. 2a). The Bragg reflection from a well-aligned CLCE should be circularly polarized with the same handedness as the helix41. To test this, we insert a λ/4 plate in the POM instrument, finding the red fibre colour enhanced with the polarizers set for right-handed circular polarization (Fig. 2b), while with left-handed polarization (Fig. 2c), the fibre is nearly invisible. In transmission without analyser (Fig. 2d), the fibre appears with the complementary blue colour (reflected red light subtracted). Between crossed polarizers with a λ plate inserted, the blue colour remains, almost unaffected by fibre rotation along (Fig. 2e) or perpendicular to (Fig. 2f) the slow axis of the λ plate (Supplementary Video 3), confirming negligible apparent birefringence from the final fibre viewed along this direction. The overall behaviour is congruent with a vertically aligned right-handed helix with submicrometre pitch41, but the top surface shows a speckle pattern with a characteristic size on the order of ~10 μm, suggesting irregularities and some variability in helix orientation near the CLCE–air boundary. Here, the precursor most likely benefited less from anisotropic deswelling due to rapid evaporation of the solvent with consequent rapid viscosity increase; the latter is a well-known effect leading to ‘skin formation’ in polymer fibre spinning48. Without the solvent, the cholesteric LCO may also have its lowest surface energy at an interface to air for normal n at the boundary, in conflict with the initial flow alignment and the effect of anisotropic deswelling, thus causing local frustration. In the fibre with green λ0 (Extended Data Fig. 3), made with a higher concentration of LC756, the surface irregularities are slightly more pronounced, although the fibre appears clearly green coloured to the naked eye.

The fibre with red retroreflection in its relaxed state provides an ideal mechanochromic response, as the full visible spectrum is available for revealing the magnitude of the elongational strain to which the fibre is subjected (Fig. 3a–k). As shown in Supplementary Videos 5 and 6, the response is immediate, fully reversible and visible to the naked eye even against a bright background. To quantify the response, we measure the reflectance spectra as well as tensile stress σxx as a function of engineering strain $${\epsilon }_{xx}={{\Delta }}{l}_{x}/{l}_{x}^{* }$$, where the Cartesian coordinate axis $$\hat{x}$$ is the fibre’s long axis, $${l}_{x}^{* }$$ is the original length and Δlx is the extension along the fibre. The results are plotted in Fig. 3m,n. At each strain level, we see a clear peak in the reflection spectrum from which we can extract λ0 (Fig. 3m). For a monodomain CLCE with a helix along $$\hat{z}$$ that is uniaxially elongated along $$\hat{x}$$, theory predicts38,40,49 that the $$\hat{z}$$ compression—and thus the mechanochromic blueshift—follows a power law $$1/{(1+{\epsilon }_{xx})}^{2/7}$$. We fit this function to our experimental λ0(ϵxx) data, i.e., retroreflection wavelength as a function of elongational strain, yielding very good results up to a strain of about 160%, after which the reduction in λ0 levels off up to 200% strain (Extended Data Fig. 4). In Fig. 3n, we visualize the mechanochromic response by plotting a subset of the experimental data against the left y axis, using circular symbols with the colour of reflection at each ϵxx value and a radius shrinking like the fibre width. The elongational stress σxx is plotted in the same way as a function of ϵxx against the right y axis.

To assess mechanical durability, we subject fibres to strain–stress cycles in a commercial mechanical testing unit, characterizing them optically before and after the tests. We find (Extended Data Fig. 5) initial Young’s moduli Y0 = 0.5 MPa for all CLCE fibres regardless of LCO, to be compared to Y0 = 0.99 MPa for a commercial rubber band. The maximum (max) strain and stress, beyond which the fibres break, differ more, decreasing from LCO1 fibres with $${\epsilon }_{xx}^{{\mathrm{max}}}=2.39$$ and $${\sigma }_{xx}^{{\mathrm{max}}}=17.4 \, {\mathrm{MPa}}$$ to the least crosslinked fibre made of LCO3 with $${\epsilon }_{xx}^{{\mathrm{max}}}=1.46$$ and $${\sigma }_{xx}^{{\mathrm{max}}}=6.81 \, {\mathrm{MPa}}$$. The rubber band has $${\epsilon }_{xx}^{{\mathrm{max}}}=6.26$$ and $${\sigma }_{xx}^{{\mathrm{max}}}=58.2 \, {\mathrm{MPa}}$$. The overall strain–stress behaviours of the CLCE fibres and rubber band are qualitatively similar (Extended Data Fig. 6), but the initial Young’s modulus of the CLCE fibres is about half as high, and the breaking strain and stress are currently substantially lower. Extrapolating the trend of fibres made of the three LCOs, we conjecture that further increase of the crosslink density should increase the strength without losing the mechanochromic response (Supplementary Discussion).

To assess the long-term durability under realistic usage conditions, we measure the reflection spectra of a pristine LCO1-derived fibre for strains up to ϵxx = 1.5 (Extended Data Fig. 7a) and then subject it to 100 cycles of ϵxx = 0 → 2 → 0. Because an important criterion for smart textiles is washability, we then run the fibre through ten full laundry cycles in a conventional washing machine. After air-drying, we subject the fibre to another 100 cycles of ϵxx = 0 → 2 → 0, after which we again measure the mechanochromic response up to ϵxx = 1.5 (Extended Data Fig. 7b). There is no change in the strain–stress curves beyond experimental variability (Extended Data Fig. 8), and the mechanochromic response is practically intact, with a λ0(ϵxx) relationship that is nearly identical to that of the pristine fibre. These data clearly show that it is realistic to use the fibres in smart textiles; hence, we end the paper with demonstrations of such applications.

## Demonstration of application potential in smart textiles

We prepare two set-ups for assessing the mechanochromic response as seen on a macroscopic scale by the naked eye under ambient illumination, in a textile context. First, we make a simple weave of ten fibres constituting the warp (vertical in Fig. 4a–d), with each fibre having its ends glued to two movable glass slides at the top and bottom of the photo. Another eleven fibres are woven up and down through the warp as the weft, and their ends are glued to another set of movable glass slides, oriented perpendicular to the first. Fibres with red and green ground-state colour appear in the weave. Supplementary Video 7 shows the weave’s dynamic response to stretching. When stretching the warp (Fig. 4b), its fibres turn green and then blue, or blue and then violet, depending on the ground-state colour. The weft fibres change less, as their deformation is minor in this setting. Upon relaxing the warp, its fibres immediately regain their original colour. If we instead stretch the weft (Fig. 4c), the same behaviour is seen with the roles inverted. When both warp and weft are stretched (Fig. 4d), all fibres are simultaneously blueshifted.

Next, we hand sew (Fig. 4i and Supplementary Video 8) a long (~1 m) CLCE fibre into the shape of the letter ‘C’ in an elastic cloth, surrounded by the letters ‘L’ and ‘E’ sewn using regular yarn. Although the fibre is tuned for red ground-state retroreflection (Extended Data Fig. 9a), it appears with a pink tone under ambient light (Fig. 4e). We attribute this to the less uniform helix orientation in this thicker fibre (Supplementary Note 1), made by reducing the mandrel rotation speed ω and translation speed v. Because ambient light effectively illuminates the fibre from all directions, domains with different helix orientations then induce Bragg reflections with different angles θ, hence varying the blueshifts, explaining the less saturated ground-state colour. The effect is much less pronounced for fibres with green ground-state retroreflection, since the high-θ reflections are in the invisible UV region. Indeed, such fibres appear to the naked eye under ambient light as clearly green in the ground state (Extended Data Fig. 10 and Supplementary Video 8.)

As the cloth is stretched in the vertical direction (Fig. 4f), parts of the ‘C’ with vertical fibre orientation blueshift, but the extent depends on the stitch orientation and tautness. The fibre colour reverts immediately and completely when the cloth is relaxed. When it is stretched horizontally (Fig. 4g), sections with taut horizontal stitches turn green. Under biaxial strain (Fig. 4h), mainly these horizontally taut sections turn green; the stretching is done by hand, yielding a biaxial strain with a weaker vertical component compared to the case in Fig. 4f. This illustrates the power of CLCE fibres in revealing complex strain patterns. As a further demonstration, we ‘program’ a fibre to monitor in-plane strains by sewing it along a 90° arc into a cloth (Supplementary Video 8 and Extended Data Fig. 10). For any uniaxial strain, a range of mechanochromic responses is seen along the arc, with each stitch showing a different colour depending on its angle with respect to the strain direction.

Based on our prior success in making 2D CLCE films of several millimetres thickness with a uniform helix by ensuring anisotropic deswelling from both opposing film surfaces29, we anticipate that uniform helix orientation can be achieved also in thick fibres if the mandrel design and material are adapted accordingly. A simpler solution, which also boosts the colour contrast against bright backgrounds, is to add a black dye to the fibre, which absorbs undesired light scattering. We do this by soaking fibres in a solution of Sudan Black (Supplementary Methods), resulting in excellent visibility of the mechanochromic response, even under ambient light over a white paper (Fig. 4j,k).

In summary, we have developed a simple procedure for making long CLCE fibres that exhibit an excellent mechanochromic response, spanning the entire visible colour spectrum, upon elongational strain up to 200%. To scale up production, longer mandrels with larger diameter can be used, and by automating the syringe translation along the mandrel, the precursor solution can be deposited with less gap between adjacent turns of the filament. Because of three factors—the fibres can be woven or sewn into regular elastic garments, our experience suggests that this will not impair user comfort, and the fibres survive long-term use as well as repeated machine washing—they can be used as smart textiles that reveal even complex strain patterns. We believe this will be particularly useful in sports clothing and wearable robotics, but it also offers ample opportunities for innovative fashion and artistic applications. Moreover, in non-wearable contexts, the CLCE fibres might serve important functions, for instance, in furniture or as a warning sign in rope (that is, a rope with incorporated CLCE fibres can signal if it is being strained to dangerous levels, or if it has been subject to strains leading to plastic deformation, such that it should be discarded).

## Methods

### Oligomer formulation

The molecular structures of the monomers used to prepare the LCOs from which the CLCE fibres were prepared are shown in Fig. 1a. The diacrylate mesogen RM257 (1,4-bis-(4-(3-acryloyloxypropyloxy)benzoyloxy)-2-methylbenzene; 1.92 g, 1 equiv.; Wilshire Technologies), the dithiol monomer EDDET (0.59 g, 1 equiv.; Sigma-Aldrich), the chiral dopant LC756 ((3R,3aS,6aS)-hexahydrofuro[3,2-b] furan-3,6-diyl bis(4-(4-((4-(acryloyloxy)butoxy) carbonyloxy) benzoyloxy)benzoate); 85 mg, 0.03 equiv.; Synthon Chemicals) and BHT (5.3 mg, 0.2 wt%; Sigma-Aldrich) were dissolved in DCM (6 ml). Then, TEA was added as a catalyst for the first-stage Michael addition reaction, and the solution was stirred at room temperature for 24 h. The LCO length decreases with increasing TEA concentration; hence, we varied its volume from 0.5 ml for LCO3 (number and weight average molar masses Mn = 12.0 kg mol–1, Mw = 19.5 kg mol–1; dispersity index = 1.63) to 1.0 ml for LCO2 (Mn = 11.5 kg mol–1, Mw = 18.9 kg mol–1; dispersity index = 1.64) to 1.6 ml for LCO1 (Mn = 10.8 kg mol–1, Mw = 17.7 kg mol–1; dispersity index = 1.63). The resulting mixture was washed thrice with 1 M aqueous HCl to remove TEA. The organic layer was washed with brine solution, dried over MgSO4 and then filtered. The solvent was evaporated using a rotary evaporator to yield the LCO. The LCO masses and dispersity index values were determined by gel permeation chromatography.

To prepare the CLCE precursor solution, the LCO was mixed in DCM with additional LC756 (1.5 to 3.15 wt% for red and green retroreflection, respectively) and a photoinitiator 2,2-dimethoxy-2-phenylacetophenone, Irgacure 651 (Sigma-Aldrich) at a concentration in the range 1–4 wt%. The overall concentration of DCM was 20 wt% for LCO1, 22 wt% for LCO2 and 25 wt% for LCO3. All chemicals were used as received.

### Fibre production

To produce the fibres, we first coated the outside of a 20 ml plastic syringe barrel acting as the mandrel with a thin layer of PVA. This was done manually using a brush dipped into a 15 wt% aqueous solution of PVA (Aldrich, Mw = 85−124 kg mol–1, 87−89% hydrolysed). The solution viscosity was high enough that the coating remained flat during drying, without breaking up due to the Rayleigh–Taylor instability. After drying (minimum 1 hour), the oligomer precursor mixture was extruded from a 10 ml syringe mounted in a Cronus syringe pump through a blunt orifice needle and collected on the mandrel, set in constant slow rotation driven by a stepper motor. The lateral translation of the syringe pump along the length of the mandrel was gently done by hand. The fibre-coated mandrel was placed at room temperature in a fume-hood for 36 hours for annealing the fibres, which were then polymerized using a UV box (Opsytec Dr. Gröbel Irradiation Chamber BSL-01) at 330–450 nm with an intensity of 200 mW cm–2 for 5 min. To harvest the fibre, the mandrel was submerged in water for 10–20 minutes, dissolving the PVA layer and allowing the entire fibre length to be easily picked up and suspended in air until dry.

### Characterization of CLCE fibre and its mechanochromic response

An Olympus BX51 polarizing optical microscope equipped with a digital camera (Olympus DP73) was used for microscopic characterization, whereas macroscopic still images and videos were acquired with a Canon EOS 100D digital SLR (single-lens reflex) camera. SEM imaging was done using a JEOL JSM-6010LA, operated in the 5–20 kV range using an in-lens secondary electron detector. Prior to SEM imaging, samples were coated with gold (~3 nm thickness) using a Quorum Q150R ES coater. The reflection spectra were obtained using unpolarized white illumination with an Andor KY328i spectroscopy platform, coupled with an Andor Newton DU920P-BXE2-DD as the detector (Oxford Instruments), connected directly to a microscope. For assessing the polarization of the reflected light, a right- or left-handed circular polarizer was placed in the beam path from the microscope to the spectrophotometer. For practical reasons, a relatively low magnification (×10) objective had to be used; hence, the area A over which the spectrum was measured was larger than the area Af of the fibre section within the measuring window. Furthermore, Af decreased upon increasing fibre elongation, due to the corresponding reduction of the fibre width. To estimate a correct value of the actual fibre reflectance R, we assumed that the reflectance value R′ given by the spectrophotometer software can be calculated as R′ = RAf/A. The normalization was done with a mirror in the sample holder. When neither mirror nor fibre was present, the reflection spectrum was a flat line at R′ ≈ 0, apart from noise (Extended Data Fig. 7d). The spectra shown in Fig. 3m were obtained by rescaling the R′ value for each data point as R = RA/Af, where A/Af was established by measuring the fraction of the image area of a photo taken at each strain level that was covered by the fibre.

Reflection spectra were plotted using Origin Pro 9.1 (OriginLab), and the central retroreflection wavelength λ0 was obtained by fitting a single-peak Lorentz function to each reflection peak in the raw data. The Warner–Terentjev λ0(ϵxx) function was programmed and fitted to the experimental data using Pro Fit 7 (Quantumsoft). To establish the strain during the λ0(ϵxx) experiments, the force F required for elongating the fibre was measured using a force gauge (Mark-10, Model M3-5), and the stress was calculated as σxx = F/a, where a is the cross-sectional area of the fibre. The latter was established by measuring the ground-state cross-sectional area using a microscope (after fibre rupture) and then calculating the area at each strain level from the elongational strain ϵxx, assuming volume conservation. For the cycled strain–stress experiments, the fibre under testing was mounted vertically in a Mark-10, F105-IMTE Advanced Test Frame, coupled with FS05-05 Tension and Compression Force Sensors (maximum loads of 2.5 N and 25 N, respectively), fixing its ends using tape. For the ten laundry cycles done after the first 100 cycles of mechanical testing, we placed the fibre in a mesh laundry bag, which we put into a conventional washing machine, running a standard 60 °C programme with detergent and ending with centrifugation. A microtome (Leica RM2200) was used to slice the CLCE fibre with 5 μm steps after it had been embedded in UV-cured NOA glue for support.

### Textile context demonstrations

Weaving and sewing experiments were both done on LCO1-derived fibres. For weaving, fibres were cut into segments of ~3 cm length and placed with about 2 mm separation in both the warp and weft. For the initial sewing experiment, a fibre of about 1 m length was coated with PVA as described in the Supplementary Methods and sewn by hand into a regular black elastic fabric for socks. After sewing was complete, the entire cloth was immersed in water for about 15 minutes to remove the excess PVA.