Some 10 minutes out from Newport, the London-bound train suddenly plunges into darkness. The clattering intensifies, the minutes tick by. The windows now serve as mirrors, shielding the traveller from the cold environment inches away.
The train seems to have slowed, but there remains a sense of speed. Few travellers realise that they are plunging some 45m below sea level along a 4.5 mile tunnel, up to the English side of the Severn Estuary.
After six minutes or so the surface world is regained and the passengers register the place merely as a landmark – the Severn Tunnel. They have no further regard for the subterranean world they have glimpsed.
In the mid-1880s, when the Great Western rail link was established between London and South Wales, the construction of the Severn Tunnel was hailed as a major feat of engineering, the longest undersea tunnel anywhere in the world.
But without the endeavours of divers, it could have been delayed for years and the construction company would probably have faced bankruptcy.
The concept of this link between South Wales, Bristol and London came about not only to permit passenger travel but to enhance the growing coal trade.
The site was at a narrowing of the Severn where the estuary was 2.25 miles wide. The tunnel would pass more than 10m below the deepest point in the river.
Construction began in March 1873. Six years on, five shafts had been sunk along the line of the tunnel and teams of miners working from these had driven about two miles of small headings.
Working below sea level, a fair volume of water was expected to seep from the rocks and have to be pumped away.
Pumps were installed in the shafts and kept pace with requirements until 18 October, when the miners intercepted a large fissure on the Welsh side of the river which discharged a tidal wave of fresh, clear water.
The Great Spring proved impossible to control. Within 24 hours the water had risen 45m. All the underground workings nearby were drowned. Amazingly, no lives were lost.
Two enormous bungs or shields were constructed to plug the source of the flow, one to block the heading which led from the base of the shaft towards the water source and a second to block the facing tunnel.
These 4m by 3m curved structures, weighing 3 tons or so apiece, would be lowered through 40m of water and guided into place by divers, braced by heavy beams set between them.
The Siebe Gorman divers, using Standard Equipment, were constrained not only by the weight of equipment and their long, heavy airhoses but had to work in darkness amid a harrowing obstacle course of gantries, platforms and abandoned tunnelling implements.
The water pressure at depth was so great that few could bear it, and significant physical exertion proved impossible.
The contractor concluded that the pressure would have to be reduced. Three enormous pumps were activated and the lowering of the shields began.
On 9 February the suction of one pump drew the lead diver, a powerful man called Alexander Lambert, so fast against the inlet pipe that it needed three strong men on a rope to pull him free!
A week later the pumps had to be switched off, the whole place flooded again and Lambert made a two-hour dive at 40m to straighten a rubber seal.
Any number of equally harrowing dives were conducted over ensuing months, but the sheer volume and pressure of water was so great that the divers simply could not get on top of the situation.
Nothing to it!
In October 1880, a year after the great flood, a way of isolating the workings from the waterflow presented itself.
Working under a head of about 10m of water, in cold water and by touch, a diver would walk up a heading through which water flowed for 300m from the bottom of the shaft.
He would step through a narrow doorway, pull up two heavy steel rails (on which trucks used to remove the rock ran), close a heavy metal door, close two large-diameter pipe valves and return the 300m to the shaft!
By now it was clear that there was only one man for the job – Alexander Lambert, 5ft 8in tall but immensely strong.
He had to be; his normal diving dress consisted of 9kg diving boots, 18kg breastplate and 27kg helmet as well as the heavy airhose. Two other divers, one at the bottom of the shaft, the other 150m along it, would be on hand to help pull the airhose forward.
Lambert, equipped only with a short iron bar, clambered over piles of debris, tools and past upturned trucks abandon-ed in the workers panic the year before.
But about 30m from his goal the friction of the airhose as it floated against the rock and around the wooden props became so strong that he could not overcome the resistance. Eventually he was forced to admit defeat.
Going back, his hose started forming itself into coils, fouling the roof supports and anything else in their way.
Patiently he disentangled it and slowly retraced his lonely routet. He returned to safety bitterly disappointed to have failed.
Chief contractor Thomas Walker, now desperate, had heard of an experimental diving apparatus belonging to a Wiltshireman, Henry Fleuss.
It was completely self contained; instead of an airhose, the diver carried a supply of compressed oxygen in a small knapsack to supply his helmet as required.
Fleuss arrived next day. His apparatus consisted of a tightly fitting waterproof facemask, connected by two rubber tubes to a flexible breathing bag or counterlung worn on the divers back.
The bag, which was connected to the cylinder of oxygen, contained a chemical which would absorb the carbon dioxide.
In at the deep end
A heavy brass helmet covered the head, but beneath this Fleuss had devised a simple but effective system for recycling the gas.
Within the tight mask, gas was inhaled via the nose and exhaled via the mouth back into the bag.
The ingenious system gave a duration of about three hours.
However, Fleuss had very little experience of diving and on none of his experimental test dives had he ever been deeper than 6m.
On 5 November 1880 the first rebreather was put to the test in the most forbidding environment imaginable.
Lambert saw Fleuss over the first part of his journey, but once into the tunnel which led towards the Great Spring, he was on his own.
Without a light and in an upright position, it was impossible to establish any sense of direction. A drainage ditch had been constructed on either side of the tunnel, so it was difficult to follow the line of the wall.
The easiest way forward was to crawl on hands and knees between the rails.
Sinking into deep mud, clambering over debris, Fleuss, understandably, soon began to falter. He finally lost his nerve and on exit stated that he would not make another attempt for 10,000.
Thomas Walker asked Fleuss to lend his apparatus to Lambert, arguing that success would present the inventor with the best possible publicity.
Lambert apparently also took some persuasion but following trials he quickly realised that the apparatus had potential.
It is worth noting that in 1880 little was known about the effects of breathing pure oxygen; it would be some years before its toxic potential was identified.
On the afternoon of 8 November Lambert began his journey into the black tunnel. Those waiting spent a tense 90 minutes before he returned.
In that time he walked and scrambled up to the door, lifted one of the steel rails and turned one of the valves as required.
However, perhaps more than a little nervous about the new equipment he was using, and with no way of knowing how long he had spent underwater, the job was still not complete.
The apparatus had clearly functioned well and Lambert was keen to complete the job. Fleuss returned to London for more oxygen and carbon dioxide absorbent, and it was two days before he set out again.
In an 80-minute dive, Lambert retraced his route to the door, removed the second rail, shut the door and turned the second valve as instructed. He returned triumphant.
Late the next day the pumps had done their work, and access was again available to the major part of the workings.
The Great Spring was eventually sealed off in early January 1881, temporarily controlled but not yet beaten.
Contained behind brickwork and other fixtures, the pressure of water on the surrounding rock strata was considerable. The workings flooded again, in October 1883.
Lambert was summoned once more. This time he failed to close the door wearing Fleuss apparatus, but managed to save the day wearing Standard Equipment.
This second flooding was sufficient warning to all concerned. It was little use containing or imprisoning the Great Spring behind so many feet of masonry, because inevitably the water would locate a weak point eventually.
The final solution was to sink a special shaft, allow the spring to drain freely to this catchment point, and install sufficient equipment to pump the water to the surface.
The engine house finally held six Cornish beam engines which worked until 1961, when they were replaced with electric pumps.
The unexpected encounter with a reservoir of fresh water beneath the land cost the company dearly and the first train did not pass through the tunnel until September 1885.
The quality of the water issuing from the Great Spring is some of the finest from any underground source, so much so that a paper mill has been located next to the pumping plant.
Today some 72 million litres are pumped daily, sufficient to supply not only the mill but also a brew-ery, the local community and the major steel works at Llanwern.Where the water comes from is still unknown.
Martyn Farrs new book, The History and Development of Cave Diving, published by Baton Wicks, is due out in November 2018.