Iron cradles on grease

The launching ways at Millwall, this third of November, were built to my own specification: a gradient of one in twelve, iron cradles on grease, and the great ship to go down sideways, stern first, into the water that had never yet held a hull of her breadth. The calculation gave a starting resistance of some two hundred tons against the winch tackle. I signed off on that figure. The figure was wrong.

At half past one she moved - a foot, perhaps two, along the ways - and then she stopped. The steam winches hauled; the chains came taut, sang, and parted. One whipped back across the crowd and killed a man on the spot, and a second died of his injuries within the day. A third chain went in the same hour. I stood on the platform above the ways and watched the largest ship ever built hold fast against everything I had specified to move her.

The failure analysis is straightforward and it is not enough. The grease was applied; the grease was squeezed out under a load of eleven thousand tons resting on the ways, and the cradles ran on dry iron. The friction coefficient I was given assumed lubrication maintained under pressure. The condition boundary was never tested at full deadweight - no way existed to test it before the launch itself. The structure performed exactly as physics required at the actual interface, and nothing like what the paper assumed at the imagined one.

The fix is specified: hydraulic rams to shove her the last thirty feet, more chains, more patience, weeks of it. But the stubborn thing is not the friction. It is that the fact was available to me beforehand - heavy loads squeeze out grease; every bearing man knows it - and it did not fit the calculation, so the calculation kept it out. The fact waited at the ways regardless. It did not care what the paper said. It did not care that two men are dead.

The rams are being procured. The ship sits on the slip tonight, holding against the tide tables and the contract dates alike. I am still at Millwall. I am still watching her refuse.