An original interferometer-based setup for measurements of length of needle-like samples is developed, and thermal expansion of o-TaS3 crystals is studied. Below the Peierls transition the temperature hysteresis of length L is observed, the width of the hysteresis loop δL/L being up to 5 · 10 −5 . The behavior of the loop is anomalous: the length changes so that it is in front of its equilibrium value. The hysteresis loop couples with that of conductivity. The sign and the value of the length hysteresis are consistent with the strain dependence of the charge-density waves (CDW) wave vector. With lowering temperature down to 100 K the CDW elastic modulus grows achieving a value comparable with the lattice Young modulus. Our results could be helpful in consideration of different systems with intrinsic superstructures.PACS Numbers: 71.45.Lr, 65.40.De Internal degrees of freedom is a feature of a random system; in principle, they can give rise to metastable size states resulting, say, to hysteresis in thermal expansion. 1 A special class form the compounds with intrinsic superstructures.Comprising two periodicities, generally incommensurate, the compounds occupy intermediate place between genuine aperiodic and truly periodic systems. 2 In these systems, such as charge-and spindensity waves (CDW and SDW), 3 Wigner cristals, superconductors in magnetic fields, 4 structurally incommensurate crystal phases, 5 the superstructure periodicities could be varied by external fields or temperature changes. The resulting metastable configurations can be reflected back onto the elastic properties and size of the underlying lattice, 3-5 though this question is still poorly understood.Quasi 1-dimensional conductors with CDW belong to a widely studied class of materials, in which intrinsic superstructure develops through the Peierls transition 6 . When electrons condense into CDW they form a deformable medium, -an electronic crystal. Deformation of the CDW affects their main static and dynamic properties and gives rise to metastability and hysteresis.The straightforward treatment of the CDW as a spring, whose strain is just applied to the crystal at the ends or via the impurities is not valid. Moreover, in the simple one-dimensional model the strains of the CDW and the crystal do not couple at all: if initially the CDW are relaxed, any change of the crystal length would not draw the CDW away from the equilibrium, i.e. give rise to a CDW deformation, 7 as it was noticed in Refs. 8-11. Similarly, once the CDW is deformed, any change of the lattice constant, c, would neither decrease nor increase the deviation of the CDW wavelength λ from the equilibrium value, λ eq . So, within this model a CDW deformation would not give rise to a length change.At the same time, the interaction of the CDW and the lattice is clearly seen from the elastic anomalies, including a drop of the Young modulus of the lattice, 8-11 Y l , up to 4%, 10 when the CDW become depinned.Mozurkewich 8 has concluded that the lattice deformation does give rise to a de...