1. Co-ordinated hunting of crabeater seals by killer whales has been described by Smith and others (1981). Leopard seals may be more important predators of crabeater seals than are killer whales (Siniff and Bengtson 1977).

2. Migration of crabeaters southward in spring and northward in autumn was recorded during the Belgica and Endurance drifts (Laws 1984:633). Ninnis later (22 January) repeats that he saw no evidence of seal migration.

3. This accords with dates for the return to breeding colonies (September–October) given by Harrison (1983).

4. Seals were driven presumably to avoid carryingcarcases. Breeding crabeaters occur in family groups of male and female with one pup. ØOritsland (1970) recorded pairs from 10 September, a few days earlier than Ninnis.

5. 23 October is within the pupping season for crabeater seals reported from Antarctic Peninsula and Enderby Land (Siniff and others 1979; Shaughnessy and Kerry 1989). At 64 lb (29 kg), the pup would have been no more than a few days old; they are bom at 20 kg and increase at 4.2 kg per day (Laws 1979, Shaughnessy and Kerry 1989).

6. Food of emperor penguins consists mostly of fish, with some squid and crustaceans. (Gales and others1990).

7. The southern giant petrel has a white phase in which the white plumage includes a few dark feathers. White phase birds are always white, irrespective of age. Thus Ninnis's white birds need not have been young. There is no earlier reference to a white skua in the report.

8. 2 November is within the pupping season reported for leopard seals in the Antarctic Peninsula region (Laws 1984: 640). Newborn leopard seal pups are estimated to weigh 35 kg (77 lb) (Hofman 1979).

9. No breeding rookeries of Addlies on ice are known.

10. Not southern right whales Balaena australis, which have no dorsal fin: more probably a large species of Balaenoptera, eg blue whale B. musculus, which enters pack ice (G. J. B. Ross, personal communication).

11. Wild (1923) described a similar fight in the February. Since leopard seals are likely to mate in late December (Siniff and others 1980), this observation seems early for courtship behaviour, and Wild's seems late.

12. Falla (1937: 25) accepted this as a record of a Cape petrel. The nearest colonies of both Cape petrels and Antarctic petrels are near Commonwealth Bay and at the Balleny Islands (Watson and others 1971).

13. Aggressive interactions between adult male and female crabeater seals have been reported in family groups while the newborn pup is still suckling and presumably before mating (Siniff and others 1979; Shaughnessy and Kerry 1989). Ninnis's observation of aggression seems late to be associated with breeding.

14. Emperor penguins breed on fast ice: no rookeries are known on drifting pack ice. Falla (1937: 38) thought it possible that this patch had broken out from a rookery, though it was more likely to mark a moulting area. Shackleton (1919: 326) noted that Ninnis had ‘discovered’ an emperor penguin rookery on the pack ice.

15. 26 November is within the pupping season reported for leopard seals in Antarctic Peninsula region (Laws 1984: 640).

16. These were possibly returning to the colony at Cape Hunter, Commonwealth Bay, where eggs are laid in late November – early December.

17. Younger age classes of Weddell seals are believed to occur in pack ice (DeMaster 1979). Ninnis recorded two Weddellsattheedgeofthepackiceon 15 and 16February, and several on 19 February. On 8 December (footnote 19) it is likely that he confused a leopard seal for a Weddell seal, and he may have done so on these occasions too.

18. Crustaceans are the main food items of snow petrels, as well as fish and cephalopods (Mougin 1975).

19. Weddell seal pups are born in September through early November in fast ice areas (DeMaster 1979). This is more likely to have been a leopard seal than a Weddell seal.

20. A photograph of this penguin is included with Ninnis's papers in the archives of the Royal Geographical Society, London (Kelly, C., in litt., 9 06 1986Google Scholar).

21. It is not clear why Ninnis thought it was a young bird. There are no external clues to their age (Harrison 1983). The ‘pearls’ may have been fish eyeballs.

22. Falla (1937: 41) considered that Ninnis provided ‘valuable information on the duration and the progress of the moult’ of emperor penguins.

23. Probably a southern giant petrel of the white phase, but not necessarily a young one (see footnote 7).

24. In the case of crabeater seals, the adult male in a family group is there to mate with the adult female; he is unlikely to be the father of the pup since the gestation period is almost 12 months and there is no evidence of a pair bond between adults.

25. This may have been a crested penguin Eudyptes sp. from the subantarctic islands, but these have yellow, not white feathers on the head.

26. ØOritsland (1977) examined stomachs of 254 crabeater seals and found they feed primarily on krill, and take small amounts of fish and squid. A crabeater seal embryo in early January seems unlikely since Bengtson and Siniff (1981) predict that implantation of the fertilized egg is delayed until early February. Presumably the embryo Ninnis found was in the reproductive tract, not the ‘stomach contents’.

27. King (1983) noted that crabeater seals moult in January. Ninnis reported two others moulting on 6 February.

28. Most of Ninnis's subsequent observations on food and stones of Addlic penguins are omitted from this text.

29. Eggs of Antarctic petrels and snow petrels hatch in mid-January. On 20 January adults would have been foraging at sea to feed chicks. The nearest colonies for both species are at the Balleny Islands and near Commonwealth Bay, approximately south-east and south-west respectively of Aurora's position.

30. Birds travelling east would have been heading toward the Balleny Islands rookeries, not those at Cape Adare, which lay southeast. On the basis of the date, these birds may have been fledged young.

31. From the direction of the Balleny Islands, where Antarctic petrels breed.

32. Presumably the straw and plywood were from Aurora's garbage.

33. Sic, presumably 26 or 27 January.

34. Possibly a white-headed petrel Pterodroma lessoni, which ranges widely in subantarctic waters.

35. Falla (1937) deduced that these were Arctic terns Sterna macrura because of their resemblance to the whitefronted tern S. striata. The latter gather over schools of Australian salmon Arripis trutta in coastal waters of New Zealand. The Maori name of the fish, kahawai, is given by New Zealand fishermen to both terns ande fish(N. G. Cheshire, in litt;, AJ. Bartle, personal communication).

36. Mollymawks with the most southerly pelagic distribution are the grey-headed albatross Diomedea chrysostoma and the black-browed albatross D. melanophris. The two species of albatross with white plumage are the royal D.epomophora and the wandering D. exulans.

37. True eels are absent from Antarctic waters and eel-like fish are uncommon. The most likely species is Notolepis coatsi which is common in Antarctic waters south to the continental shelf (R. Williams, in litt.).

38. There is no indication of how close to Macquarie Island Aurora passed - neither Mill (1916), Shackleton (1919) nor Wordie (1921) discuss that part of the journey. The only ‘land birds’ that venture to sea are shags Phalacrocorax atriceps purpurascens, kelp gulls Larus dominicanus or Antarctic terns Sterna vittata, but they are mostly coastal and do not form large flocks. If Ninnis's ‘land birds’ were in fact sea birds, they may have been shearwaters, petrels or prions.

39. Presumably a rockhopper penguin Eudyptes chrysocome or royal penguin E. schlegeli.