1. Comment:

This manuscript is a timely (perhaps slightly overdue) contribution ... potential to be a highlight paper ...  potential impact is immediate and multifaceted ... represents a genuine step forward for transparent, FAIR thermochronology"

Response:

We thank the reviewer for the positive comments!

2. Comment:

"Fission track datasets are currently reported as small data tables."

You might consider modifying this sentence to acknowledge that FT datasets are sometimes accompanied by supplementary data files.

Response:

Reviewer 2 did not like this sentence either. We will change it to clarify that the small size of the data tables refers to the fact that they are stored as ascii files, and do not contain the raw image data.

3. Comment:

Line 35: "...and accuracy of fission track data (Nachtergaele and De Grave, 2020)."

It would strengthen this section to include several more recent and relevant works on automated or AI-based fission-track identification and analysis.

Li et al. (2022); Boone et al. (2023); Ren et al. (2023); and Boone et al. (2025).

Response:

Agreed. The requested references will be added to the revised manuscript.

4. Comment:

Lines 111-113: "It is one of the reasons why fission track analysis is often done relative to age standards: observer bias does not have to be a problem provided that it is consistent between grains, and between samples."

This sentence may oversimplify the challenge. In practice, unknown samples with a high proportion of non-track features (etch artefacts, scratches, or inclusions) can exhibit strong observer dependence. In such cases, even if a consistent bias exists for standards, it may not translate reliably to unknowns. The authors may wish to clarify that observer bias can be controlled by relative calibration but not necessarily eliminated, particularly in complex or low-quality samples.

Response:

Would the following change of words help? "observer bias does not have to be a problem provided that it is consistent between samples and standards."

5. Comment:

Line 162: "Everyone counted tracks in the same ROIs."

A clearer phrasing could be:

"Everyone analyzed the features in the same ROIs."

Response:

Agreed. We thank the reviewer for the useful suggestion.

6. Comment:

Lines 167–169: "Furthermore, the vast majority of students counted fewer tracks than PV did, often many fewer, and on two occasions someone counted no tracks at all. PV's count is always above the students' median and nearly always above the upper quartile (Figure 3)."

This section raises interesting questions: Why do the student results vary so widely? Were the students' counts affected by differences in motivation or training?

It would help to specify whether this exercise was voluntary or part of an assessed class assignment. The distinction matters, since motivation strongly influences the reliability of such crowdsourcing results. The analogy to Galton’s experiment is apt, but unlike his participants, students under evaluation may behave differently if the task affects grades or workload.

Response:

The geochron@home exercise was part of an assessed exercise. However, marks were only awarded to the mathematical aspects of the exercise and not to the accuracy of the results, because I did not want to give the students an incentive to cherry pick data. However, it appears that a significant number of the students have taken advantage of this. I have called this behaviour 'vandalism' in the manuscript (see the next point). Reviewer 2 appreciated that the vandalised data are included in the study. Indeed, they are useful, because they provide a good test case for future crowd-sourcing efforts. The geochron@home exercise will be repeated in 2026 and the statistical methods introduced in the paper will be used to identify vandalism and penalise it. This deterrent should greatly improve the consistency of the crowd-sourced results.

7. Comment:

Lines 178–179:"Students near the bottom of Table 1 have undercounted the samples by such a large degree that their work can be qualified as vandalism."

The term vandalism may be too strong or ambiguous for this (although I would have written exactly the same word). If this statement is retained, the authors might clarify what they mean (data of no interpretive value?).

An interesting, related (though off-topic) point concerns p-hacking and selective reporting. For example, in a recent review of Turkish FT literature by us, approximately 40% of published FT ages report chi-squared p-values near 1.0. This raises questions about data selection and reporting bias. It may be worth discussing how FAIR and transparent data repositories, such as geochron@home, could help identify and mitigate such practices by exposing raw data and full grain distributions.

Response:

This is an interesting point. Vandalism and cherry picking are two sides of the same coin. Cherry picking happens when people have an incentive to produce "perfect" data. Cherry picking can be avoided by accepting the fact that not all data are perfect. However, the acceptance of "ugly" data can lead to flawed data being published or, in this particularly extreme scenario, vandalism.

8. Comment:

Figure 4a:

For clarity, consider revising the axis labels to:

"Track counts in grain 23” and “Track counts in grain 25."

Response:

Agreed.

9. Comment:

General comment on transparency and data culture:

There is an inherent relationship between quantity and quality, or between "how many" and "which" grains are analyzed. This balance directly influences data integrity across all research fields, including thermochronology. The manuscript correctly notes that the interpretation of FT datasets requires a high degree of trust between the analyst and the data user.

The first author has previously discussed the number of grains required for meaningful statistics in a paper, and this concept has probably been wrongly adopted as a benchmark by many reviewers and editors. However, this has led some researchers to "squeeze" rocks for marginally suitable grains just to meet a numerical threshold, which may inadvertently degrade data quality.

Transparency mechanisms such as those implemented in geochron@home can alleviate this problem. By providing open access to raw images and full counting data, journals and reviewers can evaluate not only the quantity but also the quality of the analyses, reducing incentives for selective reporting or oversampling.

Response:

The first author pleads guilty. The paper mentioned by the reviewer is "How many grains are needed for a provenance study" (Vermeesch, 2004, EPSL). It shows that at least 117 grains are needed to ensure, with a probability of 95%, that a sample contains at least one grain from any fraction greater than 5% of a detrital population. This message is easily misunderstood as meaning that only samples containing >117 dates are worthy of publication. In reality, it is obviously much better to have 50 trustworthy dates than 117 shoddy ones. As the reviewer points out, geochron@home will hopefully help prevent excessive "squeezing of the rock".

 

Pieter Vermeesch, on behalf of all the authors

1. Comment:

"This is a clearly structured, well-written work ... significant step toward improving reproducibility ... the cultural shift it may encourage ... could foster an atmosphere in which FT work is routinely assessed from its raw components upward ... I recommend acceptance with minor revisions and clarifications."

Response:

We thank the reviewer for the positive comments, which are in line with those from Reviewer 1.

2. Comment:

"It is also worth acknowledging the broader questions that may arise once such a framework is adopted. For example, to what extent can a reviewer be expected to interrogate the raw imagery, and how differing interpretations should be handled if a reviewer disagrees with elements of the dataset. These issues lie beyond the scope of the current manuscript but will inevitably require community-level discussion as FAIR-style reporting and review become more widely adopted."

Response:

It will indeed be interesting to see how this will pan out. It would be useful to carry out a larger intercomparison study of fission track experts to quantify the acceptable spread in their counts. This could lead to reviewer guidelines, so that disagreements between reviewers and authors are only considered if they exceed a particular cutoff. Tamer et al. (2025, doi:10.5194/gchron-7-45-2025) has recently completed such an exercise and observed 10% dispersion in the track density estimates obtained on the same images by multiple analysts.

3. Comment:

Line 2: "Fission track datasets are currently reported as small data tables."

To maintain neutrality, I recommend removing "small." Current guidelines (e.g., Kohn et al., 2024) encourage submission of comprehensive raw data tables as supplementary material.

Response:

Reviewer 1 made a similar remark. Irrespective of file size, data tables contain limited amounts of information compared to the raw imagery from which the numerical data are derived. The text will be rewritten to better reflect this point.

4. Comment: 

Line 34: "Ongoing developments in artificial intelligence generate further opportunities to improve the throughput and accuracy of fission track data (Nachtergaele and De Grave, 2020)."

Recommend adding a reference "Boone et al. (2025)" (under review)

Response:

I knew that this paper had been submitted but did not know that a preprint was available. A citation will be added to the revised manuscript.

5. Comment:

Line 35: "But despite the richness of the digital datasets produced by these novel tools, fission track data are still reported as small summary tables."

As above, recommend removing "small".

Response:

We will follow the reviewer's suggestion.

 

6. Comment:

Line 61: "Note that the raw microscope images are generally not stored as .jpeg files but in uncompressed czi (for Zen Blue), .tif (for Fission Track Studio) or nd2 (for Nikon/TrackFlow) formats."

Regarding raw microscope images and file formats: Fission Track Studio (both V3 and V4) can generate .jpg images by default, and these can be imported, viewed and analysed in FastTracks.

Response:

That is good to know. The statement about .tif images applies to legacy data and will be updated accordingly.

7. Comment:

Line 64-65: "For example, if a user has already counted fission tracks in Fission Track Studio, they can then store those results in a .json file at the 'sample' level directory. Because Fission Track Studio stores results in an .xml format, a second conversion script was created to translate those results into an equivalent .json format."

Since FTS .xml files contain a wide range of metadata (e.g., ROIs, auto/manual counts, c-axis, Dpar, uncertain feature flags, grain notes), it would be useful to clarify whether the json format implemented here can also accommodate this information. Although this is not the main focus of the manuscript, a brief outline of what metadata geochron@home can store would help potential users understand the platform's capacity for full transparent reporting.

Response:

The .json schema has undergone several revisions in the past and will continue to be updated in the future to accommodate additional data. For example, whilst reverse engineering FastTracks' .xml files for automated track counts, we noticed that these are stored in a different format as FastTracks' manual counts. We also noticed .xml tags for dpar and dper, which have not yet been ported to .json, but will be in the future. These examples illustrate the appealing future proofness of human-readable database formats like .json and .xml.

8. Comment

Line 58-69 (section 2-2)

This section provides important information for users preparing imagery to upload. I recommend adding a figure illustrating the directory structure and file layering. If available, providing suggested or minimum numbers of reflected-light and transmitted-light slices would also be helpful. Alternatively, presenting the settings used at University College London would offer practical guidance.

Response:

The directory structure is exceedingly simple. It consists of:

(i) a set of folders, each containing sequentially numbered .jpg or .png images, along with a rois.json file containing the vertices of the regions of interest;

(ii) a single count.json file per sample with the coordinates of the etch pits, as well as any horizontally confined fission tracks in each grain.

This information will be added to the revised manuscript.

9. Comment:

Line 80: "Depending on the permissions granted to the user by the administrator, ... In contrast, 'superusers' are allowed to define their own regions of interest."

A brief explanation of what qualifies a user to become a 'superuser' would add clarity.

Response:

Who becomes a superuser is up to whoever manages the geochron@home server. The reviewer is free to request superuser status on the UCL server by emailing its administrator (i.e. the undersigned). Of course, anyone can be a superuser on their own geochron@home server!

10. Comment:

Also note a typo: "microsocope" -> "microscope"

Thank you!

11. Comment:

Line 91: Fig. 1

The ROIs differ in size between GR, MR and GT, MT (smaller in mica images MR and MT), yet the scale bars are identical across four images. Recommend revising and re-exporting the figure to avoid potential confusion.

Response:

This must be an optical illusion. The ROIs are exactly the same size. Perhaps the appearance of different ROI size is caused by the fact that the mica images are offset vertically from the grain images?

12. Comment:

Line 106: Check the expression ρ1/ρ1 = 0.94.

Response.

Well spotted. Thank you!

13. Comment:

Line 133: "Given the right permissions (assigned by an administrator), users can build tutorial pages by annotating features in fission track images."

Similar to the comment for Line 80, outlining the levels of user permissions (e.g., entry-level, intermediate, superuser) and their associated capabilities would be beneficial. This could be incorporated into the proposed structural diagram for Section 2.2.

Response:

We are unsure if an additional diagram is necessary, but the revised manuscript will definitely address the reviewer's comment in words.

14. Comment:

Line 178-179 (Section 6 Crowd-sourcing fission track data)

The authors present an interesting experiment reflecting the strength of geochron@home. The decision to include the full suite of student dataset is appreciated. The severe undercounting observed in the lower third is indeed striking. It may also be helpful to interview the top-performing students (for example, the top third) which parts of the tutorial helped them identify features correctly. Their feedback could provide useful guidance for improving the tutorial.

Response:

We are pleased that the reviewer appreciates that we practice what we preach: the paper shows all the data, warts and all. Back when the geochron@home exercise was set, the format of Table 1 did not yet exist. We are planning to re-run the crowd-sourcing experiment again this coming year. Table 1 will be shown to the new cohort of students before they start the exercise. We anticipate that this will significantly improve their results.

15. Comment:

Line 209-210: First, a 'quiz' will be added to the tutorial pages to ensure that novice users do not count the tails but the etch pits of fission tracks.

Recommend adding an exercise that trains users to click the "centre" of each etch pit. This would reinforce good counting practice and would also provide cleaner, more consistent data for future machine-learning development, especially for high-density grains.

Response:

This is a good idea. Unfortunately, it won't be possbile to add new features to geochron@home in the foreseeable future due to a lack of time and funding.

16. Comment: 

Line 241-242: "Using geochron@home, a single microscope can serve multiple users and make fission-track analysis more affordable."

It is not entirely clear how this conclusion follows from the information provided. It would be helpful if the authors could outline a practical example or workflow that illustrates this point.

Response:

The reviewer could send her fission track slides to London, and have them scanned on our microscope. We can upload the images to the server, and the reviewer can count the fission tracks from the comfort of her office in Australia. Thus, two labs could share one microscope.

17. Comment:

General Comments on geochron@home

I completed the training section and reviewed all PV and AC counts in Section 3. The issue of distinguishing multiple track openings located very close to one another remains challenging in the current counting interface. While I agreed with nearly all of the authors' counts, I would have identified an additional 1–3 tracks in several grains where openings appear adjacent to those already marked. This is not a flaw of the manuscript but highlights a broader point: as FAIR-based reviewing becomes more common, the community will need to define how deeply a reviewer is expected to re-examine primary imagery and what threshold constitutes grounds for rejecting or disputing raw counts. These are future considerations rather than criticisms, but acknowledging them may further strengthen the discussion.

Response:

This is an interesting point that was already alluded to before. Building on the work of Tamer et al. (2025), which the reviewer was involved in, it would be useful to invite a dozen or so experienced fission trackers (including the two reviewers) to count the same images. This exercise would serve two purposes:

(i) it would quantify the acceptable range of biases in ways that weren't possible before.

(ii) its results could be used to build a novel type of tutorial, teaching novice users which features are certainly tracks (i.e. those identified by all analysts), and which are unclear (i.e. those for which some analysts disagree).

Pieter Vermeesch on behalf of all the authors