Map 1 (From Adrian Tuddenham's article in "Weigh-house". No21. Autumn 1997).

Site 5 is marked on a following map.

Site 1.
The southern side of the pound above lock 1 has a massive freestone wall; the geology (Fullers earth) and location on the contour lines and dimensions are all correct, but there are no signs of a nearby reservoir or second caisson or exit tunnel (which would heve been very long on such gently sloping ground).


Site 2 This is the 'traditional' site. Torrens obviously believes, as with Clew, that the caisson is here, as he points to 4 different pieces of evidence, all pointing to the same spot: (we now believe of course, that this is the site of Caisson1!)

1. The 1796 Map & notes Smith-Cary map shows the the course of the caisson to the SW of the late inclined plane. (Hard to see!)

2. The Geological Evidence. The caisson was cut in the (approx.) 45 feet of Fullers Earth (with its characteristic small oyster fossil of Praexogyra acuminata) in this area. From a comparison with the Geological Map ("Frome") one can see that this starts at approx. the summit of the canal here, that is the 240ft contour.(The top wall must have been a few feet above this of course ..."so much higher than the upper canal as to contain a height of water just sufficient to cover the Caisson when opposite the upper level" (Farey 1806))

3. "Evidence from the 1810 map .. is crucial." (Torrens).(Shown as Map 2 here) The canal basin that probably served the top caisson can clearly be seen to the SW of Caisson House (CH on the map). Map 2 should be compared with Map 3

4. James Tunstall's "Rambles about Bath and its neighbourhood" (1876) mentions Combe Hay ("A lovely spot though art, Combehay" (sic)) and that the site of the caisson was near to Caisson House and marked with a chestnut tree. "The drop was 60feet and the walls are believed to be still perfect as when filled up".This tree sits atop a massive artifical mound which clearly is hiding something. Mitchell (1874) stated that " The flag-pole (see map 1 and 3) opposite Caisson House marks the site of the ill-fated Caisson"- "This cannot be as the effective depth of the caisson and its distance from the entry point (approx. on the 240ft contour line (250ft is marked on the map) do not match. However the most likely explanation is that the flag-pole marked the entrance to the second caisson!" (e-mail from Adrian) Clew originally thought that the site must be somewhere between the present remains of locks 5 and 6 in front of Caisson House. However he apparently now concedes that the chestnut tree marks the spot. (See map 1) Writing in the November 1999 'Weigh House", Clew obviously still believes that the caisson is here. The nearby hillock which was once partly excavated by Mr R Bilby and his son (for a school project). ..."They ... found many rusty nails and numerous stone chippings which seemingly came from 'dressing' stone blocks for further use. We wondered why this was done, not realising that Ray Bilby had discovered part of the caisson site." (p10).

Not unreasonably, Clew states that probably much of the stone work was taken away and used in the construction of the locks, rather than in the construction of Caisson House. Another piece of evidence is that the field by site 2 was later named "Caisson Field" and was purchased by the SCC on 24/02/1797. Why would they have needed to have purchased this land if not for the caisson(s)? The land around site 4 was not purchased until 27/05/1812, well after the demise of the caisson. In answer to this Adrian points out that it was the practice to lease the land before buying it. Good though all this evidence is (1) Why haven't we found anything during excavations, and (2) the Railway (see map) was built very close to site 2, and yet there is no mention of such a caisson in any such depositions!

Nov 2003. Excavations took place in the region of the fore-bay of lock 5 using a mechanical excavator ; nothing certain was found although it's possible that the filled-in tunnel and old tunnel floor was uncovered.


Adrian also gives another, alternative , explanation for the deviation in the line of locks between locks 5 & 6.
Previous texts have assumed that this strange 'kink' is because it needed to by-pass the caisson site. Adrian's explanation is the kink is a relic of the building of the locks here which needed to go
under the inclined plane which was still operational. He also opines that if the caisson site was nearby, the locks (lock 5), would NOT have been built close to it because of possible leakage/ subsidence into the former well.
Please visit:
Kinky link

Site 3
The pound by lock 6 is larger and deeper than the others, perhaps this was the caisson site, opened-up so to speak to form a pound. However, there are no signs of exit tunnels and the heights of the top and bottom of the excavation do not appear suitable either.

Site 4 (Engine wood site)
Excavations by BIAS many years ago at the site showed lots of old brickwork but it was soon realised that this was the site of the pumping station!.. ( and thus probably not the caisson). Besides, the 'new cut' i.e. the branch canal or Top Reach, was built in 1801 - that is , according to Torrens, after the closure of the caisson.

Please note that since I wrote the piece below it's probably true to say that the pendulum has swung again in the opposite direction and that the site for caisson 1is now believed to be as Clew & Torrens first postulated. ...mmm ... read on ...

However ... New Evidence:

    The top reach (i.e. the feeder to the former pumping station) would not have needed to have been built when the caisson was in operation because the canal coming from Paulton had not got beyond Dunkerton at that time. There surely would only have been a short length of trial canal to test out the caisson.
    This negates the above paragraph.

  • The canal ("top reach') now existing at the pumping engine site is full width (14ft+) for some distance. As the engine's consumption of coal would have been about 1 boatload per fortnight, a narrow reach would have sufficed because two boats would never need to pass.
    At Crofton, there isn't even a feeder canal, the engineman had to fetch the coal up in a wheelbarrow from the main line below. Why would a full sized canal reach have been built to Engine Wood at a time when the S.C.C. was strapped for cash? Or did it already exist because it was part of the caisson experiment?

  • Sutcliffe's Estimates for the cost of replacing the inclined plane with locks:
    "Cutting a Draw from the tail of the last Lock to the Engine pit,two feet wide, three feet high, and arching it over where necessary - £564 10 0
    Sinking the Engine pit 30 [yards and] walling it where necessary, and one of the Caisson pits will serve for it, as far as it is sunk." £65. 0 0

    The depth of the engine pit is known to have been 135ft but they only estimated for sinking 90ft of that depth. The rest, 45ft, was the internal depth of the caisson; so it looks as though the pumping engine well was sunk through the floor of the caisson.

  • The ledge below the present engine site is approx. 43-45 feet lower - just about the drop in Weldon's design and consistent with the view thatthe tunnel here would have been 'robbed-out' for stone and then allowed to collapse.

    (Somewhat crude it is admitted)
    The boundary between the two distinct sounds from the geophone followed a curving path on the western side, exactly where the backing rock face would have been behind the wall of the caisson if the stonework had been robbed-out. This was approx. 80ft long i.e. the length of the caisson!

    On 29/9/99 Another survey was undertaken with an unbiased observer.
    The results can be seen at:
    Adrian's seismic survey -data.

    PS. On 2/10/99 I too went to the site and listened with Adrian. I can confirm the reduction in the bass reverbration as one crosses the 'west' wall. It is interesting to note, as Adrian says, that tree roots in this area often grow from walls. There are trees marking the north, south, east & west "walls". Those to the east and west 'bulge slightly from the mid-line - just like Weldon's design.
    However a danger here is that it way be the tree roots themselves that are contributing to this anomoly.

    Please visit this next site. I've taken pictures (Sept 1999) of this site and very crudely, have attempted to add various annotations.
    Pictures of the pump-house site
    (Jan 2000)
    Adrian writes:
    There is further news from the Engine Site. We have uncovered the capping of the disused well and found (with the geophone) that there is probably a solid connection a long way underground between the well and the engine foundations.
    There are several maps which show the caisson (or all three caissons) in caisson field. When these were scanned for a presentation it was found that they all needed to be turned to get north at the top. - All by the same angle!
    "It appears they were all copied from the same original map which pre-dates the actual building of the caisson and so cannot be taken as evidence for its final construction site."
    However, as Mr. F Pole has observed (Oct 1999), an alternative explanation might be that all the cartographers used Magnetic North raher than True North.

    If the caisson was in Caisson Field it would have to have been built at right angles to the contour lines so as to avoid an extensive and tortuous lower exit tunnel. This would need a right-angled turning basin built in the top canal to allow boats to sweep round into the caisson entrance. There is no sign that such a basin was excavated back into the hillside so it has always been assumed the canal was turned in such a way that the caisson stuck out over the gentle drop (where we excavated and found nothing). If this had been the case, the caisson would have been embanked for much of its height and length and would have been relatively immune to geological pressures. The fact that it leaked and started to collapse, virtually proves that it could not have been at Caisson Field.
  • If it were at Engine Wood, it would have been built on the natural 'nose' of a projecting contour; this would give the shortest exit tunnel and cheapest structure, without any bends either top or bottom. It would also make it susceptible to geological pressure on the uphill side, where it was cut back into a rock face.

    The exit tunnel would have been 45ft below the upper canal level. There is a plateau about 43 - 45ft lower down the Engine Wood hillside. To cross the valley would then have needed a small aqueduct, similar to the one at Dunkerton Hollow. The next caisson (which was started and abandoned) would have had to have been further along that 45ft-below-summit-level contour. If the contour 'nose' theory is correct,we would expect to find it at another sharp turn in the contour.
    Quarter of a mile down the valley, at a 'nose' in the contour, is another excavation. It is now occupied by Glen Cottage but is marked on the map as 'old quarry'. The material it could have supplied as a quarry would only have been useable as backfill, if at all. The site is about the size of a caisson.

    "The pumping engines were planned as a pair but the first one to be built was on the eastern (downhill) side of the (obviously man-made) plateau. This does not make sense in ordinary engineering terms, because it would have needed a lot more preparatory work than if it had been on the western, uphill, side. However, if the remains of the caisson were being used as a foundation, the damaged uphill side would take time to stabilise after the water was removed and replaced with infill, so the downhill side would have to be the first side to be built on."

    I am indebted to a SCC member, Adrian, for this exciting information. Most of the above detail has been taken from e'mail correspondence between the two of us. I believe he has put up an excellent case although I'm not quite convinced (yet) about the geological evidence. I also feel that a company that was willing to start work on the second and perhaps third, caisson when the first was still unproven, would have built the foundations at least, for the aqueduct about which we have no plans, accounts or financial estimates.
    Also (from Torrens) comes this piece of information:
    "In 13th Nov.1795 the Committee had advertised for...proposals... for undertaking about 30 yards of deep cutting and driving a tunnel 30 yards in length by 13ft high by 10 1/2 feet wide to an intended caisson lock." (my emphasis)
    The upper reach is much longer than 30 yards and is not in a deep cutting; nor is there a tunnel (unless of course it refers to the exit tunnel!)
    The Committee must therefore have known of the location of the caisson site and thus one would have supposed that the (?next) map of 1796 would be accurate.

    I also wonder whether a cash-sensitive committee would have countenanced the extra cost involved in taking the canal, via a short aqueduct, across a valley when there was no need.

    Also, as Halse R. (1999) has pointed out: The land around site 4 was not purchased until 27/05/1812, well after the demise of the caisson. In answer to this, Adrian points out that it was the practice to lease the land before buying it.

    However, Adrian's final piece of evidence is rather good. He has scaled and overlain maps of the area with his curved wall and the pump plans. The effect is startling!
    Adrian's overlays at Engine Wood

Site 5. The 'old reservoir' above Caisson House This can be clearly seen in Map 2

I have no other information about this site.


BUT (March 2005). I've just heard that a dig at the Pump House site has revealed NO evidence of the caisson!.

Adrian has also done an excellent
History of Rowley Bottom and the Combe Hay Locks (with other links)
Map 3.


Torrens. H.(1975) "The Somersetshire Coal Canal Caisson Lock". (BIAS)
I am indebted to Simon Jones for a copy of this document.
Clew. K.R. "The Somersetshire Coal Canal and Railways".2nd ed. Pub.: Head. 1982.

Dutens. "Notice sur les Elevators".
Sorry no more info. on this important and, perhaps, previously untapped, source. I have asked for more detail.
A. Tuddenham ( countless e'mails!)

I am indebted to Jackie Dixon for the translation from the French.

If anyone can assist with re-reading the French to ensure the translation is accurate please contact me.

Please contact me by E'mail:rtjstevens@btopenworld.com