Buenconsejo et al. measure first and later generation products from the OH oxidation of benzylalcohol. They find similar products as in earlier studies. While those were done at high NO levels, they determine the yields of the main products here at three different NO levels. While yields are constant for benzaldehyde, they find opposite trends for hydroxybenzylalcohol (HBA) and hydroxyl oxopentenal with NO level. They also measured the SOA yield from the oxidation of the two main products, benzaldehyde and hydroxybenzylalcohol. Since they find for the latter a much higher SOA yield as well as a higher product yield, the authors conclude that oxidation of hydroxybenzylalcohol is the dominant pathway to SOA formation from benzylalcohol.

The paper reports very few experiments and measurement uncertainties are fairly large. I have serious doubts, these results are robust. For example, in the SOA experiments 850 μg/m3 of HBA were added to the chamber but only 40 μg/m3 reacted (Table 1, Figure 5). There is no reason why not more HBA should react after 50 min reaction time. It is also not clear, why HCreact does not start at zero in the beginning of the experiment in both cases. There is also SOA available at the beginning of the HBA experiment or the y-axis scale is wrong. Could it be that HBA is lost to the wall and there is an equilibrium between wall and gas phase such that any further reaction of HBA cannot be observed? This would mean that not only the OH exposure would be wrong, but also the SOA-yields. From these experiments as presented here one cannot conclude anything.

Regarding the gas phase experiments the authors report only one experiment for each of the 3 NO levels. The experiment without NO has a yield of (69 ± 44)%, which overlaps largely with the uncertainty range of the other experiments. The uncertainty of the no-NO experiment is also much larger than for the others. How can the authors be so confident that HBA-yield is much higher in the no-NO experiment, which is the main result of this paper? 

While the benzaldehyde yield is similar to other studies, HBA yield is much higher. Could this also be an issue of calibration. As I understand the uncertainties also include a potentially systematic shift, i.e. if the sensitivity with respect to the used proxy differs in one or the other direction. Thus, the HBA yield could be systematically lower or higher. For example in the case of no-NO, the yield of all compounds given in Table 3 ranges from 35% to 158%. In Table 3 even not all first generation ring opening products are included as well as compounds that could not be detected by their method. This could indicate that their measured yields could be strongly biased high. It is also known that compounds like benzylalcohol and its oxidation products may be prone to wall loss. How did the authors correct for this?

This paper has serious deficiencies and needs major improvements. Furthermore, there are many flaws that should also be corrected as given below.

Other comments:

Eq. 2: this equation is a bit confusing. You use branching ratio and yield interchangeably. Branching ratios are usually ratios of two reaction paths and not fractions as it should be in this context. In Table 3 yields are reported and not branching ratios.

Line 145: not all ring-opening products are later generation products, see Figure 4.

Figure 1: for consistency I suggest to use yields for axis title and figure title. The numbers shown are fractions and not ratios.

Figure 2: In the phenol path only CH2OH not methanol can leave, to form phenol. In the benzaldhyde path OH is not removed. HO2 is formed from H-abstraction. In the HBA path OH does not add but oxygen abstracts H-atom, or oxygen adds and HO2 leaves.

Line 167: it says that yield of benzaldehyde channel should not be NO-dependent. Give reference.

Line 169: peroxy benzaldehyde: how does this radical look like? I do not think this is the correct name. Is it not something like a-peroxy-benzylalcohol?

Line 170: formation of hydroperoxide benzaldehyde would mean that an H-atom leaves the intermediate instead of HO2. Is this a feasible pathway? How can nitrate benzaldehyde form?

Line 176: Figure 3 does not show benzaldehyde oxidation, rather benzylalcohol oxidation. Furthermore, Schwantes does not show phenol formation from alkoxy benzene, it is nitrophenol or nitrosophenol. The mechanism given in Namysl is at high temperature and does not apply here.

Line 179: the adduct cannot decompose to phenol +methanol rather phenol + CH2OH radical.

Line 180: I do not see a trend in Table 1 and Figure 1.

Line 189: In Table F1 you show oxopentanal not hydroxy oxopentenal. What is correct?

Line 193: in Figure 4 no hydroxy oxopentenal is formed. As mentioned before, this compound is also not listed in Table F1.

Figure 4: oxopropanoic acid is not correct. Oxygen must be on all carbons. Hydroxyacetaldhyde is also not correct, it must be glyoxal.

Line 245: HBA rate constant is even lower than that for BA, contrary to your claim that OH-groups increase the rate by factor 4-8. Please comment.

Table F1: several chemical structures and their names do not agree, e.g. Butenedial, hydroxyoxopropenal, oxopropanoic acid. In some cases also observed m/z and reagent ion are strange. Check carefully