Considering the serious criticisms of the two reviewers on the ACPD version, with reviewer #3 even stating that he could not recommend the manuscript for publication, I feel that both reviewers should have a chance to evaluate the revised manuscript. Also additional reviewers should be given this chance.
I have some serious comments myself:
1. The very large discrepancy between the SR-XRF RDI and the ICP-MS filter data for the elements Na and Mg, with the SR-XRF data being a factor 2.5 larger than the ICP-MS data, points to a very serious problem. One source of uncertainty, which the authors fail to discuss in their manuscript, is self-absorption of the emitted X-rays for the lightest elements (especially Na and Mg). Very few details are given on the calibration of the SR-XRF measurements. On page 7, lines 18-35, it is said that use was made of multi-element standards. It is not stated how these were prepared, but reference is given to three earlier published papers. In the most recent one (i.e., Richard et al., 2010), it is said that multi-element standards were prepared from multi-element solutions using a high-resolution printing process. I presume that the standards were prepared in the same manner for the current work. How did the standards look like? Was a homogeneous thin layer created? If so, how thick was it and was correction made for self-absorption of the emitted X-rays within the layer? Or did the layer consist of individual particles of larger size than the aerosol particles in the RDI samples. In any case, X-ray self-absorption effects in general and those caused by particle size effects should not be neglected for the light elements (Na, Mg) in X-ray emission spectrometric (XRS) analysis. For a discussion of self-absorption because of the particle size effects in XRS analysis, I refer to a paper by Formenti et al. (JGR, 115, D01203, 2010) and references cited therein. If no correction for self-absorption is made in XRS, the data for the lightest elements are underestimated. It is very strange that the SR-XRF data for Na and Mg are much larger than the ICP-MS data. Could it be that the X-ray attenuation for Na and Mg in the standards was much larger than in the RDI samples and that no correction was made for this? In any case, I wonder how accurate the SR-XRF calibration factors for the elements Na and Mg are. An alternative way of calibrating is through making use of Micromatter standards (www.micromatter.com). These consist of a homogeneous thin layer of known thickness and composition, so that the correction for X-ray self-absorption can easily be calculated.
2. Considering what I wrote above, self-absorption effects cannot be neglected at all as a source of uncertainty on page 10. It also follows that the estimated total uncertainties for the lightest elements (Na, Mg) should be substantially larger than those for an element as Fe; as a consequence, Table S3 needs to be modified.
3. Figure S3b and Table S2 indicate that the SR-XRF RDI data for V are much lower than the ICP-MS filter data for this element. Yet, on page 6, bottom paragraph, of the Supplement the authors suggest that the ICP-MS data for V could be underestimated. But if this would indeed be the case, the discrepancy between the SR-XRF data and correct ICP-MS data would even become larger. Was there no problem of (serious) spectral interference for V in the SR-XRF analysis? Could it not be that the V Kα peak area was underestimated for the RDI samples?

The correction for X-ray self-absorption led to very large changes for the SR-XRF data of the elements Na-Si, with for Si still a decrease by a factor of about 2 in the finest RDI size fraction. As a consequence, a correction for self-absorption is definitely also required for several elements beyond Si, certainly for P, S and Cl, and perhaps even for K and Ca. I suspect that such correction may decrease the SR-XRF data for P and S by 25-30%. Furthermore, the uncertainty that results from the correction for self-absorption should not be limited to the elements Na-Si, but should also be incorporated for the subsequent elements. Thus, Table S3 needs to be modified and the total uncertainties should be increased for the elements from P on. Also the data in Table S1 will be affected.

The authors have properly addressed the X-ray self-absorption for the elements beyond Si.
However, a minor correction is needed in the Supplement. In the heading of the Table on page 8, “Table S1” should be replaced by “Table S2”.