Respirable Dust Measurements in the Brick Manufacturing Industry

I have with great interest read the paper by de Vocht et al. (2009), which compared respirable dust samples collected in the brick-manufacturing industry using both a Higgins–Dewell type cyclone and a dual-fraction IOM sampler. The size separation in the dual-fraction IOM sampler is carried out by a foam inserted into the nozzle of the IOM sampler. I would like to comment on some of the regression analyses presented and, what I consider to be, a lack of analysis of their data. There is something strange with the model for logtransformed concentration data as it is presented in Fig. 2 and Table 4. The least-squares regression line for the log-transformed data plotted in Figure 2 goes through the origin with a slope (b) of 0.86 (as stated in the figure caption). However, this b value is not presented in Table 4, which contains slopes for all the four models with the log-transformed data. Neither does the line go through the data presented in Figure 2, which makes it a strange result of a regression analysis. In the paper, no arguments were presented for why the regression model should go through the origin of the log-transformed data. Even if a proportional regression model might be valid for the un-transformed respirable concentration data for the IOM dual fraction filter and cyclone filter, respectively, (i.e. through the origin of the un-transformed data),