General Comments

I’m very excited to see work on improving cosmogenic 36Cl processing methods. This paper clearly presents a method for cosmogenic chlorine sample processing that splits the measurement of stable Cl from the cosmogenic measurement to improve various aspects of sample processing. There are a number of key concerns in the community around chlorine sample processing and this paper is a step towards solving some of these. Overall, the authors did an excellent job of communicating everything clearly and the paper was easy to follow. I have a couple of key points, but overall I recommend this paper for publication after minor revisions. This is a much-needed step forward. 

Specific Comments

I think the most important point of this paper is the ability to use significantly less isotopically enriched carrier (spike) for each chlorine sample. The importance of this point in this field cannot be overstated. Spike price has been increasing and there is the possibility of increasingly dwindling supplies in the future since it is not a commonly created reagent. This issue has come up regularly at conferences and is a concern for everyone in the field. The reduction of the carrier needed for each sample offers a significant improvement. This point could be highlighted even more, if desired.

The paper discusses only using 37Cl-enriched carrier and explains why this is preferable to the 35Cl-enriched carrier. However, many labs are using 35Cl-enriched carrier instead for a number of reasons (cheaper, more readily available), so I feel that this technique could be applicable to a broader range of labs more quickly if you address the issue of whether or not this technique would be possible with 35Cl-enriched carrier instead and if any changes might be needed (or when to proceed with caution). Obviously, this is not the ideal situation, but given how many labs use this carrier already, it would be nice to see it addressed directly.

 This is a very practical paper, so I will bring up one more practical point: adding additional steps at the AMS facility can be challenging when you do not have an AMS nearby. I love the fact that measuring Cl in advance would help processing be more exact, but it will also add time to the processing as well as an extra trip to the accelerator which will add considerable time to each sample batch (and potentially more cost for additional measurements?), although this will affect some labs more than others (e.g. those where samples are delivered in infrequent large batches). This is not a huge point and I think the advantages are worth the additional time in this case, but maybe worth mentioning?

Here are a number of smaller points/questions:

• Are you using a syringe filter after the Ba step?
• When crushing samples, why only use the 125-250micron sized fraction for the bulk composition? There is the potential for some fractions to crush more easily than others, which would differentiate the different grain size fractions in terms of composition. Perhaps not a problem with your mostly-homogeneous basalts in this case, but perhaps for others?
• Spoon splitting? As you state, this will matter more for some types of samples than others. Did you mix up the sample before dipping in the spoon? There are also other variations of splitting methods (such as cone and quarter) that are between a grab sample with a spoon and a riffle splitter that might preserve the idea of a more homogenous sample with less cleaning/concern (when it doesn’t matter as much).
• Can you reuse the cathodes at all?
• Line 333: (e.g., when analyzing feldspar mineral separates) – remove the word "feldspar". No need to specify since any kind of mineral separate will generally be smaller amounts.
• Line 337 - I really appreciate the discussion of other (as-of-yet unsuccessful) methods. I wish more papers had this. 

This technical note describes a non-standard sample preparation and measurement method for cosmogenic 36Cl determinations by AMS. This method is highly useful for measuring Cl more precisely and saving significant amounts of the isotopically enriched chlorine spike that is typically used for isotope dilution 36Cl AMS measurements, but becomes rare and expensive. Comparison of samples treated with the standard and the non-standard methods convincingly validate the non-routine protocol. This work is therefore very helpful and might be beneficial for other 36Cl users and AMS facilities. The manuscript is very easy to read and understand, and the figures are of high quality. I list a few comments and questions below, but can highly recommend the publication of this paper.

A significant drawback of the presented non-standard method is that the sample preparation workload and waiting time for the final AMS results is close to doubled. Thus, the main argument of reducing the expensive spike might cancel out by the costs for additional work time and the additional AMS measurement. In my opinion, this should be included in the discussion, as it is probably the main reason why this method has not been adapted by more users so far.

In this context I was wondering if a comprise could consist in combining stages 2 and 3 of the new workflow by spiking the sample grains for 36Cl measurements, as in the standard method, but with less spike and by bulking with the Br carrier, as in the non-standard method. I guess this should be possible at least for low-Cl samples (for higher-Cl samples the isotope dilution would not be sufficient), e.g. after a preliminary determination of approximate Cl concentrations for a set of samples from the same location. Has this been considered or tested?

Throughout the manuscript, the minimum 36Cl/Cl ratios necessary for precise 36Cl concentration determinations was not given much importance. I think this aspect could be more considered.

• Line 55: “A consistent sample mass (usually ~10-20 g of milled rock for whole-rock silicates or ~5-10 g of isolated mineral separates)”: Theses masses should be adjusted based on age estimates, altitude and compositions of the samples and might need to be substantially higher to obtain 36Cl/Cl ratios significantly above the blanc, e.g. for very young surfaces (see e.g. our 36Cl dating of Last Glacial and last-millennium glacial surfaces in Charton et al., 2022 https://doi.org/10.1016/j.quascirev.2022.107461 ).
• Lines 211-212: the amount of rock sample is said to be adjusted “to ensure consistent total Cl among all targets in each analytical batch”. Is the impact of the sample mass on the 36Cl/Cl neglected in this calculation? I guess that optimizing amounts of Cl and 36Cl can be in conflict for some low-36Cl samples? In this context it would be helpful to have a better idea of the 36Cl blank contributions in the presented sample 36Cl determinations. According to my back-of-the-envelope estimates they are on the order of 1-6%. (BTW, this is also missing for the Cl determinations.)
• Line 282 (and complementing my previous questions): it would be interesting to know if Cl and 36Cl blank corrections are generally lower, similar or higher with the non-standard method.

Line 295: I might have missed it, but couldn’t find the explanation why the uncertainties in the concentrations of the homogenized samples are that much smaller than in the non-homogenized samples. Please clarify if not done yet.

Lines 318-219: “Rock sample chlorine concentrations can thus be determined with high precision”. It might be interesting to add a sentence about how this can impact the precision of the final 36Cl concentrations.

Finally, I was surprised that John Stone and Keith Fifield are not mentioned, as they have been using this method for many years and have willingly been sharing detailed information about it. If this work benefitted from their experience, I guess it would be appropriate to acknowledge them.

Irene Schimmelpfennig