Firstly, we would like to thank Dr. Martin for her helpful comments on our paper, as well as her thoughts on improvements. We note our response to the 3 main comments and technical points below:

Reviewer Point 1

1. In their Figure 2, for the 1820s, why does the black line, Bellinghausen, extend north of the region of red-line region of interest? Also, this reviewer cannot distinguish the dotted from the dot-dash line. Perhaps you should use a separate color to distinguish these two. Also, Parkinson (1990, Table I) defines the Weddell as extending to 30°E, to include all of Cook’s observations. Do you want to extend it to match her discussion?

Authors’ Response:

We will improve the line marking/colouring in the final figure. Our chosen Weddell Sea area, shown by the black line in Figure 1, is only indicative. Bellinghausen’s track basically determined our northern extent, but this wobbled back and forward around 60S. We will note this in the revision. The point regarding Cook’s data will be covered in the response to Point 3 below.

Reviewer Point 2

1. For December in the Weddell, which has multiple regions of ice and open water, how do we know what the ice edge represents?

Question for authors: for this date, where would you put the ice edge in the Weddell? Following your discussion in your Section 2.5, how do you handle observations of the multiple ice edges as shown in this figure?

Authors’ Response:

The reviewer raises the very good point here that in places the sea-ice in the Weddell Sea is not continuous, but includes polynyas and eastward extensions at its northern end that have open water regions to the south. In our paper we are dealing with the northern limit of sea-ice, on the assumption that even if open water existed further south this would not be detectable by the explorers. We can make this clearer in the final manuscript.

Reviewer Point 3

1. Following Parkinson, consider addition of Cook’s data to your analysis. As stated above, the Supplementary Material in Martin and others (2022) contains Cook’s observations of icebergs and the ice edge in an Excel format, which can be easily uploaded into Google Earth. Since Cook’s first traverse in December 1772 delineates the sea-ice peninsula shown in Fig 3, this would make a excellent addition to the authors’ work.

Parkinson concludes that “ the evidence from this preliminary study suggests that it is unlikely that

a thorough study will yield a strong and consistent Little Ice Age signal from the sea ice of the Southern Ocean, at least for the 1770–1850 period.” Do the authors agree with this? Your analysis combined with Parkinson’s suggests that for the 1770–1850 period, the Dec-Jan-Feb ice cover in the Weddell is enhanced.

Given that Cook’s data is now formatted and easily available, its addition to their paper would consume a negligible amount of time, enhance their citation of Parkinson (1990), extend their analysis fifty years back in time to 1772-73, give them more data east of the prime meridian in the December Weddell, and allow them to comment on the historic regional climate. For these reasons, I hope the authors extend their analysis through addition of Cook’s data.

Authors’ Response:

We thank Dr. Martin for making us aware of her 2022 paper, which we were not aware of at the time of writing the paper. The inclusion of Cook’s data would add an additional and rewarding element to our paper. However, after careful consideration, while intending to include discussion of the Martin et al. (2022) sea-ice data and its implications in the discussion of a revised manuscript, we wish to leave the current paper’s main data presentation alone. We feel that the current paper is looking at the decadal variability of a specific 30 year period and that adding data from 50 years previously would distort the paper’s focus. In addition, the corresponding author is currently working on a circumpolar comparison of the Cook and Bellinghausen circumnavigations of both their iceberg and sea-ice records which is planned for submission later in 2023. We feel this planned paper is the best place for the comparison of the Cook data with our current work, and will place the comparison on a different geographical and temporal setting.

Reviewer’s Technical Corrections:

General: Would like to have every line numbered, not just 5, 10, 15, 20…

Line 73: The following sentence needs work.“More specifically, records utilised comprise ship logbooks, meteorological registers, charts and journals, recorded during [from the various] Antarctic expeditions.”

Line 89: Rewrite the end of the first sentence, which currently reads “region of study. This sector is presented in Fig. 2.” As “…region of study (Fig. 2).” This saves five words.

Line 96: point out that Parkinson (1990), in her Little Ice Age paper, also examined Cook’s Weddell data from 1772-73, as well as data from the early nineteenth century.

Line 145: Time and date of observation. Are you sure that all of your ships are using civil time and the civil calendar for the dates? Or, are some of them still using naval time (noon-to-noon for days)?

Line 160: How accurate in terms of km or nm is the longitude determined from a chronometer?

Line 338: Minor edit in Acknowledgements. “The authors would like to thank Théo…”

Figures: note with the possible exception of Figure 3, the figures are necessary.

Authors’ Response:

We are happy to adjust the manuscript lines 73, 89, 96 and 338 as suggested by the reviewer. Thank you for pointing these out. Line 145: The ships were using civil time and positions are usually at midday on the date given. This is noted in the metafile on the UKPDC entry. Line 160: We intend to add the accuracy estimate for the longitude value as noted in the Martin etal. (2022) paper of 6-19 km in the revision of the manuscript. Figure 3 could be left out, as much of the information is in the text, but it does serve a useful purpose in illustrating the monthly and decadal variability in observations. We would like to retain Figure 3.

Firstly, we would like to thank the reviewer for their generally positive impression of our paper and the helpful comments that will enable us to produce an improved revision. We will check the manuscript so as to remove unnecessary repetition and over-long sentences in the revised version. We note our response to the 3 main comments and technical points below:

Reviewer Point 1

 The authors mainly rely on the ship records to recreate past SIE and have talked about the inaccuracies to some extent (sec. 2.3). To clearly understand the underlying inaccuracies, the authors should state the uncertainties in the reference British Antarctic Survey digital map of the study region. Were any checks in digitization accuracy applied for the region during 19th century? Also, it would be useful to let readers know about the landmark’s used in the study to ascertain that the landmarks have not changed in the 200 years period. The errors in the route are evaluated ‘relatively’ and all the ship measurements were taken using chronometer. What was the reason that only ship Tula required significant adjustment? How big these corrections were? Stating this here might be of interest to researchers handling similar datasets.

Authors’ Response:

The BAS digital map was assumed to be highly accurate for the coastal regions being considered, with any errors under 1 km. As far as we are aware there will have been no significant changes in the landmarks chosen by the ships for navigation purposes since the nineteenth century, as these were islands or other coastal landmarks, and not glaciers which might have changed position over time. As noted in the reply to Reviewer 1, we intend to add the accuracy estimate for the chronometer-determined longitude value as noted in the Martin et al. (2022) paper of 6-19 km in the revision of the manuscript. Almost all of the vessels whose journals were used were naval vessels, with similar levels of positional accuracy independent of nation. The Tula was a sealing vessel from the 1830s, and so may have had less reason for the accuracy required of naval captains. There was no indication from the journal as to whether the errors were instrument-related or just human calculation error. The 1820s sealing cruises were often manned by crews with Napoleonic war-time experience and so probably higher levels of experience and did not require adjustments to their positions.

Reviewer Point 2

 The manuscript presents a significant amount of detail in observed data preparation which is commendable. However, pictorial illustrations of the records seem to be missing. For e.g., Fig. 2 should be updated with a background imagery (possibly optical satellite imagery) to reflect the different sea ice features present in the study region. The manuscript aims to ‘estimate summer sea ice extent in the Weddell Sea during the early nineteenth century’ and a comparative trend analysis has been discussed and presented. However, a concluding illustration is missing. The reviewer suggests adding a map (along with a proper base map) showing an appropriate combination of 1. the latitudinal locations of ice edges identified, 2. corresponding satellite dataset for selected timestamp, and 3. a final quantified SIE extent (with total estimated area in km2).

Authors’ Response:

Reviewer 1 also implieded the possibility of a summative figure. We take this on board and will add a Figure 6 something along the lines suggested by the reviewer here, although we note the difficulty of dealing with sea-ice positional data from across the whole summer season clearly. We propose to use a January image, as this is the month with the most data, and across each decade. We also think this is a better solution that adding to the complexity of Figure 2.

Reviewer Point 3

 Figure 3a seems unnecessary. The caption for Figure 3 can be more descriptive. For e.g., Nineteenth-century observations of what? The number of total observations recorded is least for March and

highest for January. However, December observations are only available for one decade i.e., 1820. Did the authors investigate any effect due to this? The authors need to state this clearly in the discussions as it affects the interpretation of multidecadal DJFM analysis specially while comparing with other studies.

Authors’ Response:

Our discussion on the best summary figure above increases our argument for retaining Figure 3. We will improve the legend (observations of ice state). We show data in detail for the different months and discuss this carefully. We will check the manuscript to ensure there is no bias introduced by the small number of December observations. The proposed January summative figure will go some way to assisting with this.

Reviewer’s Technical Corrections:

Line 200: delete ‘and will be outlined below ‘. It is repetitive with the successive sentence.

Fig 2: All the labels are not mentioned in the caption.

Line 192-193: ‘The reconstruction … Fig. 2’ can be simply written as ‘The reconstructed voyages’ data is shown in Fig. 2.’

Authors’ Response:

We are happy to adjust line 200 as suggested.

All the labels are already mentioned in the caption of Figure 2. However, Reviewer 1 commented that the line styles were not always clear so will improve the details of the figure to resolve this.

We are happy to adjust line 192-3 as suggested.