Dear Lola Brégonzio-Rozier et al.,

Regarding Table 2 (Page 22517, lines 6-7:): Please add a clear statement in the text that the differences between your results and the literature data are not significant. The reader should have no chance to misinterpret your data.

Regarding Table 4: Please add a footnote to the table that the range of primary yield values from your work is partially caused by the variability in the initial NOx levels.

Regarding aerosol mass yields:

Demonstrate the almost negligible particle wall losses e.g. by showing the plot with the dilution correction in the supplement. Refer to this in section 3.1.3.

Show a control experiment in the supplement to define the background SOA formation level. Since most of your data was obtained from experiments without seed you should show a control run without seed or both.

Include our findings on wall losses of gaseous condensable species in your discussion in section 3.1.3. See section 4.1.1 in Saathoff et al., Atmos. Chem. Phys., 9, 1551–1577, 2009 for potential gas phase and particle losses in a metal (aluminum) wall chamber. Such losses are hard to quantify and depend on the chamber mixing time scale, wall surface type and the interaction potential of the lost gas. Hence the loss rates may be much higher for e.g. low volatile acids than for those compounds studied by Wang et. al. 2011. Furthermore, the direct gas phase losses can increase in relative importance for decreasing SOA concentrations and can therefore impact the methacrolein and isoprene experiments stronger than those done with a-pinene at higher concentrations (Wang et al., 2011). These losses can therefore strongly affect the yield calculation and also impact the observed nucleation behaviour. For example, we have seen more than 95% direct SOA mass losses via the gas phase in our 84 m3 aluminium chamber for pinene precursor concentrations in the 100-200 ppt range (not yet published).

A reasonable uncertainty analysis of the yield data (best estimate) should be added to the manuscript identifying the major contributions. This should lead to reasonable error bars in Figures 3 & 7 and Table 1.