Thank you for your answers, clarifications and additional results. Please insert all this (corrections, more details on the varve counting - chronology, the comparison between OxCal and Bacon...) in a corrected manuscript.
=> major revision

Dear authors,
Many thanks for the revised manuscript. You took into consideration both reviewers' comments and this improved the original text. There are still some "technical to minor" changes that need to be addressed before I'll be pleased to accept your paper for publication.

• " For instance, we will mention that expected uncertainty of a 35μg C sample is ± 0.0113 F14C, which translates to 149 14C years if the sample is 4000 years old, but would be smaller for younger samples and larger for older samples." This is a good idea but I'm not able to see it in the revised manuscript. Please consider to add it somewhere (even in Figure 1 caption)
• Figure 1 caption: please complete the caption by adding the range of ages this figure is drawn for
• I missed the point on "if age results do not agree well, the youngest age is most likely to be correct (assuming no contamination by modern carbon)." raised by reviewer #1. There is no rationale behind that assertion. It is possible to contaminate both young and old samples in lab and young and old samples are both subject to rewording in the core. => please remove or provide me with serious arguments.
• Each sample also spends less time on the AMS when introduced as gas rather than graphite. => to be corrected. This is not true in the current state. It is true that we avoid graphitization step that in this way reduce the treatment time. If sample is introduced through EA-GIS interface, it requires a bit more presence of the AMS pilot than if introduced through the solid interface and if it is introduced through cracking GIS, it means much more time for the pilot around the machine as he has to adjust source tuning for each sample. You toned down a bit your recommendation by adding that point but arguing that the time in the machine is shorter is not at the same level. By the way, it is not clear to me which interface you use (EA or cracking). Furthermore, during the chemical step, working with ultra-small sample is much more time-consuming than with large samples. We have to very careful not to lose any fragment during each rinsing step and be sure it takes time and we definitively prefer to work with large samples. You can say as short as you did, that working with gas source spares time. That's not true. We remove one step but we spend much more time on the remaining steps (in chemistry and on the machine itself). Please adapt.
The message is that now with MICADAS we are able to run small samples and it is better to run a small sample than a big one that would not be well constrained (because it covers more than one varve or because it is bulk) and thanks of that it is now possible to get independent dating for more levels that it previously was. That's sufficient. I'm definitively not sure you need to push argument on time or on cost.


Thank you for submitting the results from your research to Geochronology

Best wishes
Christine Hatté

Dear Dr. Zander,
thank you for this revised manuscript and for taking into account the latest comments. It is my pleasure to accept it for publication.
thank you for considering Geochronology to publish your result.

all the very best
Christine

Dear Dr Zander,

Many thanks for choosing to publish with Geochronology.

With best wishes,

Philippa