The full stratigraphic sequence:
7. Sandy loam grading up to modern soil at top of cliff.
6. Upper slope breccia layer, with sharp limestone fragments mixed with glacial and glaciofluvial materials. Indistinct -- maybe 50cms thick.
5. Possibly redeposited or remobilised till with signs of glaciofluvial lenses. Melt-out till? Up to 1.5m thick.
4. Apparently fresh till full of erratic cobbles and boulders, mostly derived from outcrops to the north. Currently inaccessible because of thick vegetation below. About 2m thick
3. Limestone slope breccia accumulated beneath and old limestone cliff -- from rockfalls and gradual accumulation with some pseudo-stratification. Up to 3.5m thick.
2. Concreted sand layer - stained red by iron oxide. Irregular beds with some interbedding of raised beach gravels. Up to 2m thick.
1. Up to 2m of concreted Patella beach gravels with abundant erratics, sand lenses and shelly beds. There are large rectangular slabs of this concreted material, and they have been separated from the underlying rock platform by solution processes and wave action.
This is a classic Ipswichian - Devensian - Holocene stratigraphic sequence such as we see all around the coasts of Pembrokeshire and also on the Gower coast in multiple localities. The erratics in the raised beach must have come from pre-existing glacial deposits in the vicinity.
I see no reason to deny that the Mumbles area was overrun by glacier ice during the LGM. I find the arguments for the "redeposition" or rearrangement of older till deposits at Rotherslade and elsewhere unconvincing, no matter what the fabric analyses may say -- it is common for till to be rearranged and redeposited during the meltout process, and for diamicton layers sometimes tens of metres thick to accumulate over a short space of time -- maybe just a few decades. I see no reason for any special pleading, and prefer to call a till a till.