Status: an unproved hypothesis. This is my preferred solution, not a demonstrated provenance. It may never be proved, particularly if the original sandstone bed has been destroyed by erosion or glaciation, or now lies inaccessible beneath the sea. The evidence-led survey of the debate is on the overview and its companion pages; this page is where I say what I actually think happened.
My preferred explanation is that the Altar Stone came from somewhere in the wider Firth of Clyde region, perhaps near Ailsa Craig — not from Ailsa Craig itself, which is a microgranite, but from a small and possibly isolated deposit of Old Red Sandstone nearby.
A small, quiet-water bed
The sandstone may have accumulated in a relatively small lake, pool or quiet-water basin during the Devonian, around 400,000,000 years ago. Such a local depositional environment could have produced exactly the kind of unusual rock the Altar Stone is: pale grey-green, fine-grained, mica-rich, cemented with baryte and calcite, and containing very little potassium feldspar. It need not have resembled the dominant sandstone facies of the surrounding region — and here the geology genuinely permits it. Rift basins of this kind are coarse at their faulted margins and become finer toward their centres, so a quiet-water interval within a broadly fluvial succession is a normal expectation rather than a special plea. (That is an inference from how such basins generally behave, not a mapped bed — I label it as such.)
The honest cost of the idea should be stated up front. The exposed Old Red Sandstone around the Clyde margins — Arran, the Ayrshire coast, Kintyre — is mostly fluvial and arkosic: a poor match for the Altar Stone on its typical beds. But as the sampling problem page sets out, the whole provenance exercise is a hunt for an atypical bed. The Altar Stone does not match the typical facies of the Orcadian Basin either, and no candidate region outside Orkney and the Anglo-Welsh Basin has been eliminated by actual sampling. The surviving British ORS outcrops are fragments of the original Devonian landscape (see the palaeogeography page); Old Red Sandstone is known to continue beneath the Firth of Clyde and the North Channel; and smaller basins may have been submerged, deeply buried or completely eroded away — the Quaternary’s 2.58 million years of repeated ice cover is ample time to remove an outcrop from the map entirely.
Even the stone’s celebrated barium is no obstacle to a western source. Baryte cementation is bed-level fluid chemistry, not a regional badge — Bevins and colleagues found baryte largely absent from the Orcadian samples they tested — and barium-rich fluid systems are well documented in the west: southwest Scotland supported working baryte mines at Muirshiel in Renfrewshire and Gasswater in Ayrshire.
The transport leg is the proven part
A glacier could then have entrained a block from this deposit and carried it south through the Irish Sea system. This is not a convenient imagined route — it is the best-attested long-distance stone-moving corridor in Britain (see the glacial transport page). Western Scottish and Hebridean erratics occur on the Gower coast. Ailsa Craig microgranite itself is a recognised indicator stone carried south through this very corridor — which is part of the appeal of a source in its neighbourhood: we already know the ice that flowed past that spot delivered rock to the Bristol Channel approaches. And the pink granite at Saunton in north Devon, probably from Gruinard Bay far to the north, shows that such journeys can span several glaciations, an erratic re-entrained by later ice.
The delivery toward Wessex would belong to the Anglian, around 478,000 ya, or to an earlier glaciation still. The stone could then have been moved only the final, much shorter distance by the people who built Stonehenge. No extraordinary 750-kilometre Neolithic expedition is required, and no equivalent monument, staging post or quarry need be missing at the far end — because there was no far end.
The anonymity problem — why nobody has found the companions
The difference between my candidate and the famous erratics is visibility. Ailsa Craig granite is beautiful, distinctive and readily noticed on a beach; it is the material traditionally used for curling stones. A weathered piece of grey-green micaceous sandstone would be far more anonymous. Unless somebody were specifically looking for its mineralogical signature, it could easily be dismissed as an ordinary grey beach boulder. Companion erratics may therefore already exist — in Irish Sea deposits, on Welsh or Bristol Channel coasts, or sitting unrecognised in museum collections — and simply never have been noticed for what they are. Nobody has yet searched.
What would prove me wrong
This is a hypothesis with testable remnants, and it should be held to them (both tests are set out on the further analysis page). Offshore borehole and core records beneath the Firth of Clyde and North Channel could be examined for Old Red Sandstone of the right character — a desk study, at existing archives, costing little. And Irish Sea erratic collections could be systematically searched for further pale green, mica-rich, baryte-bearing sandstone blocks. A match in either place would transform the case. Equally, I should say what would count against it: a sustained, systematic search of the erratic collections that finds nothing would genuinely weaken the hypothesis, and hafnium-isotope work that pointed firmly and exclusively to the far northeast would weaken it further. An idea that cannot say what would count against it is not worth holding.
None of this proves that the Altar Stone came from near Ailsa Craig. The precise source bed may no longer exist, and a unique match may never be found. But the region has not been categorically eliminated; the geology permits a lost or submerged source, and the glacial transport route is already demonstrated by other Scottish erratics carried to the very coasts where the stone needed to arrive. For me, it remains the most economical and satisfying solution presently available.