The following is based upon my reading of Hammer WR, Collinson JW & Ryan WJ (1990).

Fossil hunting is something we occasionally do as a family, and a few conditions have to be fulfilled or we won't set out. The weather mustn't be cold or rainy, travelling time will be no more than a couple of hours, a pleasant restaurant is a welcome accessory (the fish and chip shop in Lyme Regis certainly qualifies), and an afternoon picnic is essential. These considerations mean, despite its attractions, our hammers and chisels will never trouble the rocks of the Fremouw Formation; strata which are way past the back of beyond.

Geography
The Fremouw is near the Beardmore Glacier in the Transantarctic Mountains. Should anybody feel like checking an atlas, it's not all that far from where the Queen Maud Mountains reach the Ross Ice Shelf. The map in the paper mentioned above makes me want to write southwest Antarctica, but all coasts of the world are north of the South Pole.

A bit of history
An early identification of Triassic vertebrates from this area was made in 1969, (p.163). A couple of localities yielded fossils, and these allowed the recognition of a Lystrosaurus Zone. A number of factors show this dates from a similar age to the Lystrosaurus Assemblage Zone of South Africa, with the most obvious being the shared, dominant genus. Further common genera and stratigraphic information confirm the fauna as deep Lower Triassic.
The fossils described in the 1990 study were from 200 metres higher up in the same formation, and are somewhat younger. This fauna has similarities with the African Cynognathus Assemblage Zone, which immediately overlies the Lystrosaurus AZ. It straddles the Lower-Middle Triassic transition in South Africa. Conveniently, the relevant Antarctic rocks in the Gordon Valley aren't beneath a glacier, and that's a contrast to most the local landscape. (I should mention the authors prefer Kannemeyeria Assemblage Zone, but state it's equivalent to the Cynognathus AZ; ie. different terminology.)
This newer assemblage was discovered late in the 1985-86 field season, and four researchers had less than two weeks to sample it. Nevertheless, they picked up almost fifty specimens, and about half are potentially assignable to the generic level. Their earlier efforts garnered a couple of hundred fossils from Lystrosaurus Zone sites.

Geology and paleoenvironment
The Fremouw Formation is around 600 metres thick and runs for about 475km, (p.165). There are three distinct subunits of strata. The lowest includes remains of Lystrosaurus and friends, and it's built from up to 125 metres of sandstone. The next member is largely composed of stone formed from silt and mud. This has a maximum thickness of 200 metres and is predominantly greenish grey in colour. Vertebrates have been found but they were unidentifiable. The upper subunit is 300 metres of volcanistic sandstone, and the Cynognathus fauna occurs towards the top of the first eight metres; for clarity, that's relatively near the base. They're in particularly resistant sandstone, and this forms a notable platform for a couple of kilometres in the Gordon Valley, with a further exposure at Fremouw Peak. That mountain provides a clue as to the nomenclatural origins for the Formation. It's the type section. This platform was the result of river action. It contains siltstones and squashed logs as well as vertebrate fossils. The animal remains are generally well preserved but disarticulated bits of skulls and lower jaws. There are also remnants of a Dicroidium flora, (p.166).
The whole Formation was deposited in a floodplain environment, where relatively slow-flowing rivers dumped debris into a basin. Sandstones accumulated in the channels while fine grained sand, silt and mud built up beyond the banks. Signs of small roots indicate some plant activity, but the squashed logs are typically in the erstwhile rivers. They were probably shipped in from further afield.

A paleo weather report
Triassic Antarctica was close to the southern polar area, but temperatures were probably rather mild. The diversity of reasonably large therapsids and amphibians doesn't suggest a freezing wasteland, and nor do nearby trees. What's known about the streams is consistent with semi-arid conditions interrupted by seasonal floods. A contribution to those may well have been made by melting snow in higher areas.

A glance at the animals
Two groups are represented in this fauna; terrestrial therapsids and water-loving amphibians. Gordon Valley is generous with temnospondyl amphibians, and they're mostly remains from large animals, (p.164). They're also generally previously unknown genera of capitosaurideans, (p.165). In one case, part of a snout came from a skull with a complete length of about a metre. As this was doubtlessly attached to a considerable body, the giants among amphibians today would've been dwarfed.
The abundance of those animals in a fauna of this age is unusual. It isn't the case for South Africa, India or South America. It is similar to the balance in some Australian localities but there doesn't seem to be a generic overlap. One explanation for this amphibian power is the southerly position of Triassic Antarctica. In some cases today, their relatives can tolerate climates too cold for reptiles. This could be useful knowledge should you fear a crocodile attack when in the bath. Keeping the water temperature below 10°C should ensure safety. However, salamanders and frogs would happily hop in.

Therapsid butchers
Being predominantly land lovers, therapsids didn't generally spend much time in rivers. Their remains presumably found their way into the water through flood action and bank erosion, and that would also explain why they tend to be less well-preserved. They took a bit of a battering during the journey.
One fragment is the middle part of a lower left jaw, which is at least very similar to Cynognathus. As no distinctions are evident it's been referred to that genus, (p.164). The fossil has a maximum length of 12cm and the corresponding skull would have measured around 30cm. This was the top-dog predator for much of the world. The front two-thirds of a right mandible, represent a different large eucynodont. Its identity is unclear but it seems to have been more derived in some respects. Its presence could imply that this fauna existed a bit later than the deposition of the Cynognathus Assemblage Zone in Karoo, but that's not necessarily the case. Derived and basal lineages can (and do) coexist.

Therapsid botanists
A small piece of an upper jaw contains the remains of a broken canine tusk. This is from a dicynodont kannemeyeriid, and it could conceivably be Kannemeyeria. (They were large herbivores, should anybody be wondering.) However, not enough is available to allow any preciser diagnosis. Assuming it wasn't built in an utterly peculiar way, the skull of its former owner would have been about 60cm long; big enough for a fair degree of personal security, as long as it didn't attempt to play with larger amphibians.
The fauna also contains a large therapsid gardener, but it isn't of kannemeyeriid proportions. There's a fairly complete right maxilla with gomphodont teeth. In size it's comparable with big diademodontids but it lacks simple postcanines at the front. There's a gap between the canine and the transverse postcanines instead. Furthermore, the bone overhangs the dentition on the lateral side. Both characteristics are more in line with traversodontids and perhaps Cricodon. However, there are many distinctions from both those taxa.

Update
Abdala et al, 2005 provide a very brief summary of this fauna on page 197. They list four lower jaw fragments of cynognathids and a large cranial fragment with teeth. They term this a diademodontid from the Gordon Valley locality. It also states that Hammer suggested a close relationship with Titanogomphodon in 1995. A study of pollen suggests an Anisian age for the fauna.

A number of Mesozoic (and post-Mesozoic) location summaries can be found at Localities.