Lots of interest in the Titanic at the moment which reminds me of a thought I had some time ago about all those rivets you see on the hulls of old ships.
I understand that a hot rivet is hammered through holes in two steel plates and as the rivet cools it pulls the two plates together.
Ok I get that, but was that good enough to make the plates watertight or did they have to use some kind of sealant/coating to stop leakage?
My observation, as a complete non-expert: those steel plates are painted – above the water line partly for appearance, but below the water line to retard corrosion of the steel plates. That paint is laid on fairly thick, and would by itself seal any small gaps between the plates.
Not sure. Boilers were made this way, and had to be water and gas tight at high pressure after many cycles changing hundreds of degrees in temperature – a much tougher requirement.
I know a mechanical engineer who is also a steam locomotive engineer, and will ask him!
There’s a bygone shipbuilding trade called caulking. The caulkers were charged with caulking those seams to make them watertight.
After they were riveted
And on a quick read, it’s not the same as the caulking that was done on wooden ships - actually driving old rope-strands or tar into the seams between planks. It seems to involve deforming the edge of a riveted plate to make it into a snugger fit with the next.
Well, don’t leave us wondering. 
How was the edge ‘deformed’, and how was it accomplished? Was it done before or after the plates were riveted together?
Wiki sez:
and mentions that the same thing was done with boilers.
One thing this leaves me wondering:
A quick check shows that both arc welding and oxy acetylene welding were developed in the late 19th century. How come they were still using riveted hull and boiler construction well past WW I? The Queen Mary, laid down in 1930, was riveted. Steam locomotive boilers were still riveted in the 1920s.
Welding wasn’t as strong as riveting.
James Burke, in his TV series Connections from the mid-'70s, discusses the first all-welded ships built in the 1940s, Liberty ships.
Burke says here (YouTube) that all-welded ships weren’t as strong as riveted ships, but they weren’t expected to last more than five years.
From the video clip that includes the Liberty ship Jeremiah O’Brien:
British shipyards were very slow to introduce welded hull construction, even in the 1940s many warship hulls were still riveted. The post-war slump meant many yards closed altogether and those that remained had cautious, ageing management who were reluctant to innovate. The hard times also lead to often bitter industrial relations and some strange trade-union demarcation dividing lines.
As to watertightness - frequently they weren’t. Single-riveted seams often loosened in rough weather, particularly in small ships like destroyers where captains might drive them hard in rough weather if they dared. The mess-decks (normally in the forepeak) could be miserable places in an Atlantic storm, with water streaming in from the loose rivets on the deck as well.
To expand on what was already said… The metallurgy and welding technology just weren’t mature enough.
One thing that riveted seams did better was stop cracks.
Here’s a rather famous photograph of a liberty ship that split clean in half because of metal stress issues in combination with few barriers to crack propagation.
I had a teacher at the maritime academy whotold the sstory of being on a liberty ship that started to crack inhalf at sea. Acording to the teacher they managed to get the two bow hooks back to the stern and the stern hook up to the bow. Then they put tension on both to hold the ship together until they could get to the nearest safe port.
I was also on a T-2 tanker that was of the complete welded hull design. Some time after WWII sshe was put in dry dock and a line was cut down the hull and butt strapped over. This way if a crack started it would travel only to the butt strap and there it would stop.
My expert says there is no caulking or gasketing in boiler riveted joints, and that the contraction of a hot rivet is an essential part of the process of sealing the joint.
Typically a pneumatic caulking operation was performed along the bottom edges of the overlapping plates, causing the steel to flow plastically into the riveted joint. No sealant required.