In a previous letter [1] it was reported that the value of the adhesive fracture energy, Gc, was dependent upon the type of substrate material which was employed for the test specimens. Even though the locus of joint failure was observed to be cohesive, approximately in the center of the adhesive layer, and the measurements were made using specimens which obeyed the requirements of linear-elastic fracture-mechanics (LEFM) tests. The results from this earlier study are given in Table I. (A single-part, heat-curing rubber-toughened epoxypaste adhesive, designed for general-purpose use, was employed to bond the following substrate materials: mild steel, aluminum alloy and CFRP (a unidirectional carbon-fiber reinforced-plastic epoxy composite). Both double cantilever beam (DCB) and tapered double cantilever beam (TDCB) specimens were employed [1, 2]. The thickness of the adhesive layer was 0.4 mm.) The results given in Table I were explained in terms of the shape and size of the plastic zone, ahead of the crack tip within the adhesive layer, changing the value of Gc. The shape and size of the plastic zone was, in turn, considered to be affected by the value of the transverse modulus, E22, of the substrates used to form the joint. However, in a recent study involving the same grade of general-purpose epoxy-paste adhesive and range of substrates [3, 4], a far more pronounced effect of the substrate material on the value of Gc was noted in the case of the CFRP joints. All tests were again valid LEFM tests, and the crack always grew cohesively, approximately through the center of the adhesive layer. (Hence, as in the previous work, any differences in the values of Gc from using the different substrates cannot be explained in terms of any changes in the interfacial adhesion as the type of substrate was varied.) The results from this second study revealed no significant differences in the values of Gc for the mild steel and aluminum alloy joints, but the value of Gc for the CFRP joints was even lower than that recorded in the first study. For the CFRP joints the value of Gc was now only 202± 23 Jm−2. (In this second study, the adhesive joints were all manufactured in the same laboratory by the same personnel, but testing was carried out by twelve different laboratories, according to a protocol [2, 4], as part of a “round-robin” series of tests on structural adhesives tested under Mode I (tensile) loading.) It was considered that the very low value of Gc recorded for the CFRP joints from this second study was far too low to be explained by the mechanics arguments advanced previously [1]. Now, in the previous study, the adhesive had always been cured under the same temperature versus time cycle of 150 ◦C for 70 min in a fan-assisted oven, whereupon the heaters were turned off and the joints allowed to cool slowly over a period of about 5 h in the oven. This relatively long curing time was employed to try to ensure that the same state of cure was experienced by the adhesive layer in all the different types of joints. Thermocouples were buried into the adhesive layer to monitor accurately the temperature of the adhesive, during the curing cycle, in combination with the various types of substrate materials. Indeed, very similar cure cycles were recorded for the different types of joints. However, it was decided to investigate the extent of cure of the epoxy adhesive in more detail by measuring the glass transition temperature, Tg, of the cured adhesive layer. This was undertaken for adhesive samples removed from the different types of joints by using differential scanning calorimetry (DSC) and employing a heating rate of 20 ◦C min−1. Samples of adhesive were therefore carefully removed from the fracture surfaces of the failed joints and these samples were then analyzed. The values of Tg were defined on the normalized heat flow endotherms as the transition half-width obtained from a second heating cycle. For the “round robin” exercise, the values of the glass transition temperatures, Tg, for the samples of adhesive taken (a) from the mild steel joints was 103.0± 0.9 ◦C, (b) from the aluminum alloy joints was 100.6± 0.6 ◦C and (c) from the CFRP joints was 87.6± 1.3 ◦C. Thus, clearly the most striking feature is the very low value for the Tg for the adhesive layer in the CFRP joints, and these joints also possessed the correspondingly very low value of Gc of 202± 23 Jm−2.