A wide range of thermoreversible gels are prepared by cooling down to ambient temperature hot aqueous polymer solutions. During the sol-gel transition, such materials may experience a volume contraction which is traditionally overlooked as rheological measurements are usually performed in geometries of constant volume. In this article, we revisit the formation of 1.5% wt. agar gels through a series of benchmark rheological experiments performed with a plate-plate geometry. We demonstrate on that particular gel of polysaccharides that the contraction associated with the sol/gel transition cannot be neglected. Indeed, imposing a constant gap width during the gelation results in the strain hardening of the sample, as evidenced by the large negative normal force that develops. Such hardening leads to the slow drift in time of the gel elastic modulus G' toward ever larger values, and thus to an erroneous estimate of G'. As an alternative, we show that imposing a constant normal force equals to zero during the gelation, instead of a constant gap width, suppresses the hardening as the decrease of the gap compensates for the sample contraction. As such, imposing a zero normal force is a more reliable way to measure the linear properties of agar gels, which we prove to work equally well with rough and smooth boundary conditions. Using normal force controlled rheology, we then investigate the impact of thermal history on 1.5% wt. agar gels. We show that neither the value of the cooling rate nor the introduction of a constant temperature stage during the cooling process influences the gel elastic properties. Instead, G' only depends on the terminal temperature reached at the end of the cooling ramp, as confirmed by direct imaging of the gel microstructure by cryoelectron microscopy. Finally, we also discuss two subtle artifacts associated with the use of duralumin plates that may interfere with the rheological measurements of agar gelation. We show that (i) the corrosion of duralumin by the aqueous solution, and (ii) the slow migration of the oil rim added around the sample to prevent evaporation, may both lead separately to a premature and artificial growth of G' that should not be misinterpreted as the formation of a pregel. The present work offers an extensive review of the technical difficulties associated with the rheology of hydrogels and paves the way for a systematic use of normal force controlled rheology to monitor nonisochoric processes. (C) 2016 The Society of Rheology.