I have covered this topic at length before -- please type in "entrainment" into the search box, and you will find many entries. Sadly, my entries are not being read by the right people -- or, if they ARE the right people, they just don't like what they read, and choose to ignore it.........
This is not the place for a treatise on glacial erosion -- that is well covered in thousands of papers and scores of text-books, including some of mine! Suffice to say that there are many different erosional processes at work beneath ice, including abrasion (in which the base of a glacier, armed with abundant stones and other debris, acts like sandpaper on overridden rock surfaces) and plastic scouring (which produces smoothed and moulded forms where the till at the base of the glacier is saturated, or where large volumes of meltwater may be flowing along the glacier bed). The effect of abrasion, when relatively large tools are involved, is the creation of deep, parallel, long striations (striae) and grooves in the rock surface.
Another process is gouging, in which large, heavy "tools" on a glacier bed are used effectively like chisels on a rock surface that is being over-ridden, creating friction cracks, chatter-marks and gouges that can be tens of centimetres deep. Many of these cracks or fractures are transverse to the direction of ice movement. Anybody who has wandered about in the Stockholm Atchipelago (as I have) will know exactly what I am talking about! The physics of these processes are complex, and are best left for another day and another geomorphologist.
What we are especially concerned about here is QUARRYING, by which large blocks of bedrock are broken away from rock surfaces or are picked up from old scree slopes or other accumulations of blocky debris that happen to be overridden by ice. Without becoming too complicated or too mathematical, we can refer to the summary below by Prof David Evans.
Essentially, what he is saying is that pressure variations on the bed of a glacier favour the breaking off of slabs or blocks of bedrock, on the basis that rock under compression (on the up-glacier sides of obstacles) is quite strong, but is much weaker -- and therefore susceptible to entrainment -- when it is under tension (on the down-glacier sides of obstacles). Where glacier ice is relatively thin, and where cavities can be created on the glacier bed, quarrying is enhanced. The process is essentially one of spalling, sheeting or pressure release -- something which geologists are familiar with, and which can sometimes lead to solid rock shattering or "bursting" with near-explosive force.
The role of water is important too --- on the compression side of obstacles, water remains in a liquid state, but on the tension side pressure is reduced, and water often freezes back onto the sole of the glacier, and in doing so incorporates or "entrains" any large blocks that have been loosened. So blocks can be dragged away and incorporated into the glacier. If the glacier is flowing smoothly, and if the glacier bed is still melting here and there, and lubricating the basal ice, the entrained boulders will remain close to the glacier bed and may not be carried very far; but if the ice thickens over time, because of a deteriorating climate or because of topographic controls, then the ice may freeze onto its bed. In other words, the glacier will change from being "warm-based" to being "cold-based." When that happens, the only way for the glacier to continue its forward momentum -- because it is being forced by accumulating ice upstream -- is by processes of internal deformation, including shearing. Thrust planes can carry big slabs and pillars of broken rock up into the body of the glacier -- and in some circumstances even up onto the glacier surface. The extent to which these blocks will be modified or smoothed during transport will vary according to the potential that there may be for abrasion or further breakages -- but now we are straying into the territory of the next chapter in this treatise.....
There now -- that wasn't so painful, was it?
The implications of all this for the Stonhenge Bluestone debate? Well, that I would not be at all surprised to find "bluestone erratics" turning up on Salisbury Plain which have come from either the north Pembrokeshire coast between Dinas and Newport, or from Carnedd Meibion Owen and Tycanol Wood, or from the northern slope of Preseli (where we find the tors of Carn Alw and Carn Goedog). I would find it very surprising indeed if erratics were to turn up from the southern slopes of Preseli. As for erratics from Carn Breseb, Carn Gyfrwy, Carn Sian, Carn Meini, and Carn Dafad-las, I would expect some to turn up on Salisbury Plain, although the circumstances for entrainment would not have been quite so favourable.
What I would expect is that erratics derived from the southern slopes of Preseli or from the crest of the ridge might be quite extensively spread across the landscape of mid- and south Pembrokeshire, after entrainment and transport during later and less extensive phases of glacial activity. We don't know how many "pulses" of late-glacial activity there might have been at the end of the Anglian glaciation, but if the Devensian Glaciation can be taken as a reasonable model, there may well have been one or more coolings equivalent of those we refer to as the Younger and Older Dryas phases in Western Europe. These coolings just MIGHT have been associated with short-lived regional glacial advances.
Quarrying (Prof David Evans)
Prog in Phys Geog 2004
Quarrying involves two separate processes: (1) the fracturing or crushing of bedrock
beneath the glacier; and (2) the entrainment of this fractured or crushed rock. Fracturing
of bedrock may take place where a glacier flowing over bedrock creates pressure
differences in the underlying rock, causing stress fields that may be sufficient to induce
rock fracture (Morland and Boulton, 1975; Morland and Morris, 1977). Fluctuations in
basal water pressure may also help to propagate bedrock fractures beneath a glacier
(Röthlisberger and Iken, 1981; Walder and Hallet, 1985; Iverson,1991a). Brepson(1979)
has successfully simulated the sliding of temperate ice over an obstacle in the
laboratory, and noted that large cavities form in the lee of obstacles, aiding quarrying.
Evacuation of rock fragments along joints in the bed is possible where localized basal
freezing occurs, for example as the result of the heat-pump effect proposed by Robin
(1976). Although Holmes (1944) originally argued that quarrying could occur beneath
both thick and thin ice, and outlined a theory based on pressure-controlled freezing of
meltwater in joints in bedrock, there is now general agreement that quarrying is
favoured beneath thin, fast-flowing ice (Hallet, 1996). Modelling studies indicate that
low effective basal pressures (0.1–1MPa) and high sliding velocities are the dominant
glaciological conditions required for quarrying because these conditions favour
extensive ice/bed separation (subglacial cavity formation)and also concentrate stresses
at points, such as the corners of bedrock ledges, where ice is in contact with the bed
(Iverson, 1991a; Hallet, 1996).