We consider the structural interpretation of Fukumura et al. (2019) to be incorrect not least because it is incompatible with any published geological mapping of the area—either previous or their own. We also consider the probability that orthopyroxene is widely present in rocks where this mineral has never been reported before to be so vanishingly small that, as it stands, this report by Fukumura et al. (2019) is not scientifically credible. The presence of talc and olivine in the Shiragayama body reported by Fukumura et al. (2019) is in agreement with previous studies (e.g. Kawahara et al., 2016). The new aspect of Fukumura et al. (2019) is the interpreted coexistence in equilibrium of the two minerals, but only textural evidence is provided for the interpreted stable coexistence. Quantitative chemical analyses of the talc and olivine would provide an important test of the proposed thermodynamic equilibrium (olivine is reported to have an Mg# of 0.98 and using an appropriate equilibrium coefficient, talc in equilibrium with this should be almost the pure Mg end-member) and until such information is provided, the assertion that the olivine-talc equilibrium assemblage exists in the Shiragayama body should be treated with some caution. In the following we give more details of the reasons for our concerns about the structural interpretation and the proposed relatively common occurrence of orthopyroxene in the Shiragayama body. The most glaring geological problem of Fukumura et al. (2019) is that their cross section of the Shiragayama body contradicts the results of both their own mapping and previous geological studies in this area (Endo et al., 2018; Kawahara et al., 2016; Shiota, 1991). In their cross section Fukumura et al. (2019) show the boundaries of the serpentinite bodies dipping to the south—locally with dips in excess of 60°—whereas previous studies and their own mapping show this boundary is broadly horizontal. If we consider structural contours for the boundary of the main serpentinite body (unit or layer 1 on Figure 1) shown by Fukumura et al. (2019), it is clear they imply the northern part of the body has a very gentle dip—actually to the north and not 20°–30° to the south as shown. The rest of the boundary to serpentinite unit (1) is shown on the map presented by Fukumura et al. (2019) as constrained between the 1 250–1 300 m topographic contours over a horizontal distance of more than 1 km. This distribution implies the boundary is broadly horizontal, not steeply dipping as shown on their cross section. As a further illustration of the problem, we examine the type of serpentinite distribution pattern implied by their cross section. For this construction, we assume that their cross section is drawn in the dip direction of the main boundaries of the large serpentinite body (unit (1) in Figure 1) and then plot the approximate implied map distribution of these boundaries (Figure 1). There is a clear discrepancy between the implied geological boundaries and the map view given both by earlier workers (Endo et al., 2018; Kawahara et al., 2016; Shiota, 1991) and that shown by Fukumura et al. (2019). Any obliquity between dip direction and the line of cross section would increase this discrepancy. The observational basis for drawing the steep dips shown by Fukumura et al. (2019) is unclear, but part of the problem may stem from the idea of a bedding schistosity—a concept where bedding and the main tectonic foliation are assumed to be parallel and which has in the past been used to interpret structures in the Sanbagawa belt (e.g. Hara, Hide, & Shiota, 1994). In mapping metamorphic domains that have undergone large ductile strain, lithological boundaries and tectonic foliations may be locally subparallel, but in general are oblique and failure to distinguish their orientations is likely to result in erroneous structural interpretations. We also note that the large fault with a throw in excess of 500 m shown in the Fukumura et al. (2019) cross section is absent from the corresponding map and there seems to be no geological evidence in support of its existence. We next examine the proposed large-scale 'duplex' structure in more detail. The main data for discussing the geological structure of an area are the outcrop patterns for different lithologies. In Figure 1 we show the distribution of different lithologies recorded in our studies including both outcrop and geologically significant loose blocks shown with crosses. These data suggest that parts of serpentinite units (5) and (4) shown by Fukumura et al. (2019) are parts of one larger unit and much of the area mapped by Fukumura et al. (2019) as serpentinite unit (4) is actually occupied by metasediments. Our data confirm that units (3) and (2) are separated by metasediments, but the western extension of unit (2) shown by Fukumura et al. (2019) crosses an area where outcrops of metasediments are observed. In addition, if the overall shallow dip of the main body is taken into account, several serpentinite bodies which are distinct and separate in map view can be linked into a single sheet, as shown in the upper cross section. The reasons for the discrepancy between previous mapping and the study of Fukumura et al. (2019) are unclear, but we note that in the area close to the main serpentinite body large blocks of serpentinite are prominent whereas outcrops and blocks of pelitic schist are strongly weathered and are easy to overlook. Publishing lithological maps that identify and distinguish outcrop and important loose blocks, such as that presented in Figure 1, would allow such disagreements to be resolved. The authors report orthopyroxene (opx) in 10 of their 21 samples (table 1 in Fukumura et al., 2019) but provide no chemical analyses or photomicrographs to support the claim. In the study reported by Kawahara et al. (2016), thin sections of more than 100 samples from the Shiragayama area were studied and we found exactly zero samples with opx. Such a discrepancy seems so unlikely as to be beyond the realm of what is reasonable or even possible as a result of chance. It is not straightforward to convincingly demonstrate the absence of something when only a small fraction of the whole is actually observed, and some quantitative examination of how the two studies compare is desirable. The approach we take is to consider the presence or absence of a mineral in a sample to be a binomial distribution and that the samples from the Shiragayama area are a random selection of the serpentinite. The question that needs to be answered is, could the two sets of data be derived from the same geological body simply by chance. This approach is not mathematically rigorous because geological sampling is not properly random. In the present case, the studies by Fukumura et al. (2019) and Kawahara et al. (2016) were conducted with similar emphases on petrological and structural geological characteristics of the region and the sampling routes reported in Fukumura et al. (2019) are also covered by the more extensive network examined by Kawahara et al. (2016). These considerations make it more likely that similar samples were examined than if the locations were random and should accentuate similarities rather than differences. In contrast, if samples reported in Fukumura et al. (2019) were selected with some particular characteristic not recognized by other workers that increased the likelihood of finding opx, this could help explain the differences. However, such selection criteria are not mentioned in Fukumura et al. (2019) and a comparison of the patterns of sampling locations does not suggest radically different sampling strategies in the two studies of Fukumura et al. (2019) and Kawahara et al. (2016). Therefore, we consider that the assumption of randomness is a reasonable approximation and any sampling biases are more strongly weighted in favor of obtaining more similar results making it more difficult to explain discrepancies, not less. Following the above discussion and to allow for chance to play a major role in the results, we employ a 5 sigma test: if the probability of finding the observed difference from the same original distribution is greater than about one in 3 million we will accept that chance may be the cause—our null hypothesis. Otherwise, we conclude that other factors, such as mineral misidentification, are the reason for the discrepancy. The result exceeds our 5 sigma criteria, we reject the null hypothesis and conclude chance is not an adequate explanation for the reported differences in mineralogy. Simple geometrical constructs such as structural contours are an important part of interpreting geological maps and they should not be neglected in metamorphic domains. Incompatibilities with earlier work in the same area or topic should be highlighted and the reasons for the differences studied: testing the reproducibility of research results is an important part of the scientific endeavor. The cross section and structural interpretation given by Fukumura et al. (2019) is not compatible with the known distribution of rock types in the area and the reported mineral assemblages are not compatible with previous more detailed studies of the same area. We respectfully suggest both should be re-examined. We are grateful to two reviewers and the Editor H. Maekawa, who provided useful comments that helped improve the clarity of this comment.