page 2, l23 you say you observe no significant excess but later you report on a 2.8 sigma excess. This is inconsistent.

The abstract has been changed to the following: "No significant excess of events over the standard model expectation is observed in the mumujj final state. An excess with a local significance of 2.8$\sigma$ is observed in the eejj mass distribution at approximately 2.1 TeV."

page 4 why is the mass of N_l not used in the hypothesis testing?

In order to allow for a more generic interpretation of the cross section limits, the search focuses on the four-object (dilepton+dijet) mass distribution. The lepton-jet-jet mass distribution, using either lepton, can then be used to classify the nature of the excess. Both distributions (either lepton 1 or lepton 2 together with the two jets) were examined in the investigation of the electron channel excess.

page 4, l43, 2nd column: acceptance needs to be defined

The "kinematic and detector acceptance" requirements, as mentioned in line 43, are defined in the paragraph beginning on line 23 on the second column of this page (page 4) to be |eta| < 2.5 for electrons and jets and |eta| < 2.4 for muons (provided one muon has |eta| < 2.1). All objects must have pT > 40 GeV, with at least 60 GeV required for the leading lepton.

Both acceptance and efficiency is 80% by coincidence? So the combined acceptance*efficiency for signal is ~64%?

The muon reconstruction efficiency is slightly higher than the electron reconstruction efficiency. As a result, the muon channel efficiency is greater than the efficiency for the electron channel. We now quote both the muon and electron channel efficiencies separately to remove the appearance of coincidence, by changing the text starting on page 4, column 2, line 48 to read "Provided that the WR boson decay satisfies acceptance requirements, the ability to reconstruct all four final-state particles is near 75% (85%) for the electron (muon) channel, with some variation due to WR boson and N masses."

Page 5, l8: can you quantify "very similar"? Why are there differences?

The selection requirements are in fact identical for emujj events with respect to eejj or mumujj events, after accounting for the fact that the control sample contains both an electron and a muon instead of two same-flavor leptons. To remove any confusion, line 8 and following now reads as "...we apply selection requirements to emujj events that parallel those applied to electron and muon channel events."

page 5, l20 you say the scale factors (SFs) are derived from simulation. Data-MC differences in the trigger and lepton efficiencies could affect these SFs. Please clarify if this considered?

The simulation response is corrected to data based on control sample studies. We add the following sentence to the last paragraph in Section 3 to this effect: "The simulation is compared to data using various control samples, and when necessary the simulation is adjusted to account for slight deviations seen with respect to data."

Fig. 2 Dark blue band looks black. Please fix. Please add bands to the legend

The blue band in Fig. 2 is now lighter. A legend has also been added to the lower portion of the figure for the systematic uncertainty bands. Still need to address description of bands.

page 6, l59-60 What other distributions did you check? Is this statement true for all possible combinations of W_R and N_l masses and for Majorana and Dirac N_l? Do you mainly refer to the different rate of the excess or also to shapes of distributions which are inconsistent. Please clarify.

Additional distributions checked include, but are not limited to, the M(ee), M(jj), and M(ejj) (for both the leading and subleading electron) distributions, as well as the pT of each final state object. A range of neutrino masses, from 100 GeV up to the M(WR) - 100 GeV, were considered. Even considering g_R < g_L, the shapes of the additional distributions are inconsistent with WR production and decay to an electron and right-handed neutrino, which would appear as a "localized excess" (as we state in the text) in these distributions.

page 6, l23, 2nd column: how many events are expected to be BG out of the 14 data events? How many signal and BG same-sign events do you expect? One same-sign event might be an excess above the SM expectation. How many same-sign events to you find in the muon channel for the same mass window?

Roughly four background events are expected with 1.8 < M(eejj) < 2.2 TeV. Assuming g_R = g_L, roughly 20 events would be expected for WR -> eN production with M(WR) = 2.1 TeV, assuming a sufficiently heavy neutrino such that acceptance and reconstruction efficiency are maximal. Half of these signal events would be expected to be same-sign. For the background, nominal SM processes, together with charge mis-ID for DY+jets (studied using simulation as well as data control samples in the Z peak region), predict roughly nearly 0.5 same-sign events from background. We find 0 same-sign events in the muon channel, although we note that the potential for charge misidentification differs for high pT electrons and muons.

page 6, l42, 2nd column: Madgraph vs Sherpa is an arbitrary choice for an uncertainty. What about pdf and scale uncertainties?

The Madgraph/Sherpa comparison was made to address theoretical shape uncertainties, including scale uncertainties. Under the assumption that scale uncertainties should not change with dielectron mass, the scale uncertainty was checked using data and simulation near the Z peak. The MC uncertainty (Madgraph/Sherpa) was found to be consistent with the data-driven comparison. We have modified the text in two places to address the additional background systematic uncertainties, which are small compared to background shape uncertainty. On page 6, the line "This overall background normalization uncertainty, together with the remaining background uncertainties, is small compared to the uncertainties determined for the background shape." now reads "This overall background normalization uncertainty is small compared to the uncertainties determined for the background shape." We later modify the paragraph on page 7, originally starting with "The total uncertainty for signal and background...". It now reads "The systematic uncertainties related to pileup, uncertainties in the proton PDFs, and initial or final state radiation are computed for the simulated background samples and are found to be small when compared to the background shape uncertainty. Additional theoretical uncertainties for the SM background processes are covered by the shape uncertainty. The total uncertainty for signal and background...".

page 7, l33, second column: it is not clear where acceptance, lepton-jet overlap, i.e. isolation and the shape is taken from. Is this based on truth-level quantities? Please clarify. As M_Nl/M_WR approaches zero truth-level studies become less and less reliable. In l43 you mention full simulation: down to what value of M_Nl/M_WR do you go? I am wondering if you can really parameterise the effects in this regime.

In order to clarify where truth-level quantities are used, in the last sentence of this paragraph we now state "Consequently, for M_Nl values other than M_Nl = 1/2 M_WR, the WR boson production cross section limits are computed using information from signal samples that do not include the simulated detector response." We check MN/MWR from 5/6 down to 1/16. The difference between generator-level and full simulation is successfully modeled by a smooth function throughout the mass range considered, and the uncertainty is parametrized for the full range of WR and N mass values we consider.