Dear editor and referee, We appreciate your valuable comments on the manuscript. Please find enclosed a revised version, where we attempted to answer most of the points raised by the referee and the editor, and also addressed a number of comments, which we received after we had posted a submitted version of the paper on arXiv. All changes in the paper (except those involving references) are highlighted in red. Before I get to the actual response, I would like to point out that the automatic text-to-pdf conversion, which was applied to the reviewer's comments by the AAS publication system produced unsatisfactory results. The PDF was unreadable if printed on paper and could only be viewed electronically. Also, it converted some of the special characters such as long dashes into some unreadable character sequences. I would kindly ask the editor to forward this feedback to the support of the AAS manuscript handling web-site. Please find below our responses to the criticism by the referee and the editor. > Editor's comments -- > In sec 2.2, I do not understand the fitting procedure where "we recover the shape of the narrow line component in a non-parametric way by solving a linear convolution problem with the regularization, which requires a smoothness of a solution". Is kernel smoothing involved? Is this a semi-parametric local regression technique similar to splines or LOESS? Please explain the procedure in full detail, possibly in an Appendix. Give references to the methodology where applicable. We decided not to extend the description to the Appendix. In fact, the extended description was published in the RCSED paper (Chilingarian et al. 2017, section 2.4.2). We did add a couple of sentences explaining the regularization technique, which is an implementation of the standard L2 Tikhonov regularization that uses the second order differential operator in order to guarantee the smoothness of the solution. > Reviewer's comments -- > 1) In the fourth paragraph of the introduction ... > So, I suggest the authors should consider separating the item (ii) the stellar dynamics and pulsar timing into two different categories. At present stage, stellar dynamics alone did not provide any black hole mass measurements, which would qualify to be called IMBH in the terminology of our paper (<100kMSun). This is because the sphere of influence is too small to be resolved beyond 1Mpc. The 47Tuc IMBH has been detected using a combination of pulsar timing with stellar dynamics, that's why we included those two techniques under the same item. Regarding the papers by Nguyen et al. 2017-2018 -- we have added the data to the scaling relation plot and a few sentences into the discussion section (see our comments below). On the other hand, the paper by Nyland et al. (2017) does not include any direct evidence for the existence of a black hole in NGC404 and it relies on the result from Nguyen et al. 2017. Perhaps the gas kinematics obtained with ALMA will allow the authors to get a solid confirmation of the central BH in NGC404 and the estimate of its mass, however, we have not found any published results presenting the analysis of ALMA data for NGC404. > 2) The authors presented the selected criteria to rule out the spurious broad line detections ... > 2.1) ... How did the authors visualize the spectra, e.g., by eyes, emission profiles, or fitting? We did not visualize the spectra in order to perform the filtering of objects where airglow lines affected the measurements because with a list of airglow lines we were able to determine ranges of redshift where at least one of them would affect at least one of the lines used for the non-parametric line fitting. For other purposes we used a specially designed web-based visualization system, which we re-used to prepare plots in figures 3--6. > 2.2) Empirical constraint that the width of the broad line component is at \sqrt(5) times larger than that of the narrow line component. Why \sqrt(5)? All the mathematics dealing with Gaussian widths is of a quadratic nature. By using sqrt(5) we aimed at the relative importance of broad-to-narrow line components of at least 5:1. We have added a sentence explaining this in the text. > 2.3) The BPT classification... We did keep both, AGN and composite spectra as it is stated in the text ("or") > 2.4) The Halpha/Hbeta Balmer decrement According to Dong et al. (2008), 99% of type-I AGN have Halpha/Hbeta < 4 -- this is the main reason why we used 4. We added a sentence explaining this. > 2.5) |vBLR - vNLR| < 3sigma NLR ... We added an explanation to the text: we wanted to reject strongly asymmetric broad line profiles because we fitted them using symmetric Gaussians. > 3) In subsection 4.4 and Figure 8, the authors discussed their IMBH masses and host galaxy properties to the recent compilations of bulge/spheroid masses of host galaxies of massive black holes... We added Nguyen+2017/8 data to Fig8 and also added a few extra data points for tidally stripped stellar systems, compact elliptical and ultracompact dwarf galaxies in order to demonstrate some examples of galaxies which are not expected and indeed do not follow the MBH-Mbulge relation. We also added a paragraph in the discussion addressing this question. > 4) In subsection 4.4 entitled Implications for co-evolution of central black holes and their host galaxies. The authors argued at the beginning of the second paragraph that... We have added "because their bulges are small" into the last sentence of the paragraph. > 5) Issues with figures 3-6: > 5.1) The contradiction of the plot and the caption Corrected > 5.2) The grey lines in the lower profiles are residuals but it is not clear for the meaning of the upper and lower bound lines? Added > 5.3) The non-parametric narrow line models (rightmost panels) are not explained sufficiently. The idea of the rightmost panels is to demonstrate a 2-components of the LOSVD recovered by our data analysis algorithm to scale in the H-alpha line. > 6) If it is possible, at the discussion of subsection 4.1, I would like to see the BPT diagram that plots all 10 targets with strong detection evidence of IMBHs We added a figure with a BPT diagram as requested by the referee and included a brief description of that figure. > 7) The authors emphasized the most dominated error budget of their IMBHs search and masses estimates method in subsection 4.3. Should the author be more explicitly for any other error sources that possibly affect their IMBH mass estimate method? If there are no others, please clarity! The current wording ("The contaminating sources likely have diverse origin.") reflects our best understanding of the contamination budget and the fact that it is uncertain. At this point we simply cannot be more specific and provide the full list of contaminating source classes and estimate precisely how they influence our mass estimates. Probably we will address this issue in the future papers once we have sufficiently large sample of IMBH candidates observed in our follow-up campaign and understand better the causes of spurious detections. --------------------------------- > II) Minor comments and suggestions: > 1) In the second paragraph of the introduction section section, I found the following sentence, â€oePopulation III stars might have formed in dense clusters in primordial density fluctuations, which could then evolve into more massive SMBH seeds by collisions and/or core collapse (Portegies Zwart et al. 2004).†does not follow the discussion flow of SMBH seeds formation mechanisms in the Early Universe. We have reviewed that part of the paper and decided to leave this sentence where it is because mentioning clusters of PopIII stars is a logical extension of the previous sentence where we mentioned PopIII for the first time. > 2) In the third paragraph of the introduction section, the authors mentioned the IMBH mass estimate for NGC 4395... The NGC4395 mass estimate from reverberation mapping presented by Petersen et al. 2005 is a factor of 3 more precise than the dynamic estimate by den Brok et al. 2015, therefore we use it throughout the paper. > 3) I may be lost in the discussion flow, but in subsections 2.1 and 2.2, the authors provided many numbers that are relative to the total optical spectra ... Here we say "about 1,000,000" in section 2.1 and then give a precise number (938,487) in secion 2.2. > 4) The authors presented the criteria of sample filtering for the feasible IMBH candidates Added a few words in section 2.1 > 5) In the second paragraph of subsection 4.1, the authors used the acronym IAU without spelling it out. Point 3 in the Style section of the AAS guidelines for authors states: "Acronyms and abbreviations should be spelled out the first time they are used unless they are common throughout the discipline." International Astronomical Union is a common acronym, which does not require spelling it out. The editor might correct it if we are wrong. > 6) In subsection 4.1 at the final paragraph ... Corrected > 7) In Figure 7, the authors show the orange star Added into the figure caption (now it's Fig.8) > 8) The last sentence of the fourth paragraph Added <1% With best regards, the authors