What Caused Titanic To Sink?
by Parks Stephenson
Just about everyone agrees that the Titanic disaster was the result of a collision between the liner and an iceberg. Below that macro view, though, there is much debate about what actually happened that caused a technological marvel like Titanic to fall so readily to a brush with a floating piece of ice. Beginning shortly after news flashed out about the disaster and continuing to modern day, an endless procession of pundits has offered up various theories about what happened. The theories range from pure (and sometimes comically absurd) speculation to scientific study but none yet have answered questions that I have been prompted to ask after a decade's worth of study of the wreck. In this article, I will describe what I have learned and how I applied that knowledge to contribute to what we can know about how and why Titanic sank.
In this article, I will explain the scientific method I used to arrive at my conclusions. Some of you may – and have – asked, "How can you claim scientific method when you are not an appropriately credentialed scientist?" My response is that scientific method requires only that an experiment be documented and reproducible and it will be up to the reader to determine if I met that criteria. I want to stress that I do not claim to prove anything beyond a shadow of doubt, but I do believe that I can show cause for challenging some of the trendy theories promoted in Titanic literature and maybe even introduce new perspectives into the mix, in the hope that each new contribution is another step closer to learning the truth.
The most commonly-accepted scenario from 1912 to 1985 was that the iceberg ripped open a 300-foot-long gash into the hull. Some did question as early as 1912 how ice could "rip open" harder steel (Senator Smith's famous question to Fifth Officer Lowe), but not enough to significantly challenge this basic scenario in the public mind. Finding the wreck 73 years later torn in two opened up new speculations. With the resurgence of Titanic interest in the 1990s-2000s – along with new exploration – came new scenarios. Artefacts from the wreck itself were recovered, providing investigators with actual physical evidence for study. Attention turned to the manner in which Titanic had been constructed, and how well the material used to make her hull performed when placed under the stress of contact with the berg. Brittle steel, weak rivets, design faults…guided by the latest expert opinion, the "story" about Titanic's loss – each one grabbing headlines – became more about material failure than human error.
To be fair, I too was caught up in this obsession with the mechanics of the collision. Based on my own experience aboard ships, I found support in the eyewitness written accounts for the notion that Titanic briefly grounded on the iceberg (a scenario first conceived just days after the disaster). In May 2001, I collaborated with David Brown on the writing of a paper that put forward that argument. The strength of our argument depended on our interpretation of survivor accounts of the collision, which naturally opened us up to criticism from others who did not share our perspective. Without hard material evidence, ours was just another argument, among many, that was based primarily on opinion and interpretation.
At about the same time, expeditions to the wreck conducted during the late 1990s yielded sonar scans and images of what was purported to be the original damage suffered during the collision. This path of investigation suffered from an unfortunate fact: the bow section buckled upon impact with the ocean floor at about the same spot where the original collision damage was located. Differentiating between iceberg and bottom-impact damage proved impossible, given the inaccessibility of the entire area to close observation. The claim that the original iceberg damage has been imaged has yet to be proven beyond reasonable doubt.
For a while, the theory of brittle steel was popular. Using the story of the T-2 tanker Schenectedy, which broke in half while fitting out in a shipyard in Oregon in January 1943, as precedent, some researchers tried to draw parallels with Titanic. The breakup of Schenectedy was not unique but dramatic enough to inspire the creation of the Ship Structure Committee after war's end, which focused research into ductile-brittle transition and how changing the molecular content of steel can increase its yield strength and lower the temperature at which that transition is experienced. The Schenectedy case study is not directly applicable to Titanic because it highlighted the failure of welded joints, of which Titanic had only a few (none of which joined the hull plating), and later impact testing (conducted in ice-water temperature) of Titanic hull steel specimens would prove that the steel used in the hull plating was tough enough to resist brittle fracture. Wreck imagery shows numerous examples of steel plating bent back onto itself with no ductile tearing. Even though some scientists have written articles claiming that brittle steel was a major factor in the sinking of Titanic, theirs was actually an opinion that was not supported by testing on any steel recovered from the wreck.
Aside from the materials testing, proponents of the brittle-steel theory also neglected to take into account two significant historical factors: 1) despite claims to the contrary, Titanic was not in cold water for an extended period of time (later collection of environmental data by Tim Maltin would show that Titanic was in the cold waters of the Labrador current for only about 4 hours before the collision); and 2) unlike the Schenectedy, the only corroborated breach in Titanic's hull (that reported by Leading Fireman Frederick Barrett in Boiler Room #6) had an operating boiler room with 100-degree temperatures on the other side of the steel. This will be given appropriate context later.
In the late 2000s, metallurgical analysis of the rivets grabbed headlines in both popular and scientific media. The claim was made that Harland & Wolff used substandard material in the making of at least some of the wrought-iron rivets used in Titanic's construction, creating plating seams of inconsistent strength and ability to withstand additional loading, such as the pressure exerted by the iceberg against the hull. In essence, the scientists who conducted the metallurgical analysis of the rivets speculated that had the shipbuilder purchased a better quality of iron bar stock from the steel mills, not as many rivets would have failed and the ship would have stayed afloat longer, buying time for rescue ships to arrive in time to save lives.
I did understand and accept the analysis of the metallurgical composition of the rivets; after all, the scientists conducting the analysis had the proper tools and were expert in their fields. Even though I wasn't a scientist, I knew from the historical record about the impurities in the iron created by the open-hearth process at the beginning of the 20th Century and the shipbuilding industry's lack of understanding of the metallurgy of steel at the molecular level until after World War II…there is no denying that Titanic's rivets were not as strong as they would be if they were made today. But as a historian and observer of the wreck, I had too many questions when it came to accepting their conclusions about the role that the rivets played in the disaster. The British shipbuilding industry had standards for steel and iron quality – even if they were somewhat uninformed when seen from a modern perspective – and material inspections were part of H&W's construction process. If the rivets had so little strength, then why are the stem and stern of the Titanic wreck, both held together by wrought-iron rivets, the most intact and recognisable parts of the wreck today, despite having slammed into the ocean floor so hard that they buried themselves as deep as 40 feet or more? Why didn't Olympic's iron-riveted seams open up when she deliberately rammed the German submarine U-103 in 1918, an impact so hard that it bent her stem 8 feet off centre? And why hasn't Britannic's structure sagged – much less collapsed – after lying on her side underwater for almost a century?
These questions were forming in my mind when I became involved in a production making an episode for National Geographic Channel's Seconds From Disaster series. An experiment was arranged by a metallurgist for that production that involved creating two steel plates and joining them together with two wrought-iron rivets – one on each side of the joint – and then bending one plate until the rivet head was forced to separate. This experiment was used to illustrate how the rivet would fail under load. Again, I am just a layman but it seemed to me that this arrangement did not accurately represent the environment in which Titanic's rivets actually operated. Causing a single rivet to fail did not seem all that complicated…just pry it up with bent plating much like you would with a crowbar. But what if there was a row of rivets, each neighbour lending strength to one another? The explanation offered by more than one scientist was that rivets would "unzip" along the seam in that instance, but would that actually happen? I may not be a metallurgist but I have had some training as a naval architect and I simply could not follow the leap in logic that the scientists were making from their analysis of the strength in just a few individual rivets (out of 6 hull rivets in their sample, only one — a decapitated rivet head — was made of iron, they did not have an intact iron rivet for comparison) to the mechanics of a seam failure in the actual ship. Ultimately, I had to believe that the shipbuilders at Harland & Wolff knew a bit more about their business than modern-day scientists were insinuating. But how could I, a layman, make that case?
An opportunity presented itself when we were successful in pitching an investigative documentary to the History Channel network on the occasion of the 100th anniversary of the disaster. Among other aspects of our investigation into the manner in which Titanic broke apart and sank, I proposed that we subject replicas of a portion of one of Titanic's plating seams to industry-standard materials testing. I had actually gotten the idea when Dan Butler showed me one of the replica rivets that he was having produced to market as a souvenir in 2012. In order to make the proposal practical, though, I had to identify exactly what should be replicated. What did we know that could help us replicate a seam damaged by the iceberg?
A fact that few ever acknowledge is that we simply don't know the exact character of the collision with the iceberg; more specifically, the damage done to the ship. In 1912, Edward Wilding calculated, based upon his interpretation of survivor observations relative to times and flooding, that the breach in the hull constituted a total area of about 12 square feet. Naturally, there would be an indefinable margin of error in any calculation based off interpretation of eyewitness statements, especially when the timeline cannot be corroborated. Sonar scans conducted during the late 1990s at first seemed to show the damage on the starboard side under the current mudline (which amazingly equalled Wilding's 12-sq-ft estimate) but similar damage concurrently imaged on the port side suggested that what the scan had found was damage caused by bottom impact, not the original collision with the iceberg. James Cameron departed from the traditional 300-foot-long gash scenario by showing the iceberg pushing in plates down the starboard side, opening seams along the way, in the animation for his 1997 movie. This reinforced the modern popular scenario of small openings down a substantial length of the bow but the exact nature of the damage was nothing more than conjecture. To add an element of complexity was the flooding reported by multiple eyewitnesses in Boiler Room #4, a compartment too far aft to have been damaged in any proposed collision scenario. The location of this flooding is so far outside the envelope that it is simply ignored by most analysts as some kind of anomaly. With so much uncertainty about the exact nature of the collision, what should we attempt to replicate for our test?
Ultimately, I decided that we should replicate the only hull breach for which we had corroboration…that of the opening witnessed by Fireman Barrett in Boiler Room #6. His description of the height of the opening corresponded with the landing between the "J" and "K" strakes, a fact that Bill Sauder was able to confirm from the original framing plan. Having identified an actual plating seam known to have suffered damage, I set about designing the test specimens. The original idea was to build two specimens, one with wrought-iron rivets, another with steel rivets, so that we could compare their relative strengths. The production company for the documentary, Lone Wolf Documentary Group, arranged for stress testing to be conducted at the University of Washington (UW) Materials Science lab under the guidance of Dr. Brian Flinn and his team. Dan Butler brought in Steve Howell at Ballard Forge in Seattle, who salvaged century-old wrought-iron from a half-sunken barge and dismantled bridge girders to be swaged (re-forged) into bar stock that would be used to make our rivets. We specifically hunted for century-old iron so that we would have slag content comparable to that used in Titanic (average 3-4% content, as compared to the 9% found by metallurgists in the decapitated head of an iron hull rivet, the only portion of a Titanic iron hull rivet available to date for analysis).
At this point, a fortuitous event occurred that profoundly changed the direction of our experiment. Simon Mills, conducting research in the Public Records Office in the UK, found and sent a copy of two memos written by the British Board of Trade's Ship Surveyor, Francis Carruthers, in February 1912. Evidently, Olympic had experienced a stormy passage in January 1912 and upon her return to port, it was found that a plating seam forward and another aft – on both sides of the hull – had flexed more than expected during the storm, loosing rivets which in turn resulted in some seeping from joints. Edward Wilding at Harland & Wolff looked at the problem and decided that the structure in those areas should be stiffened in order to reduce the magnitude of the flexing. This stiffening required the installation of a 1" thick reinforcing strap to be fitted over the landing for the affected seams. My heart jumped when I read the affected area forward: "In way of the No. 6 boiler room & extending three frame spaces forward of the W.T. bulkhead at the forward end of the boiler room, viz.:- from frame 63 to 81 at the landing of the J & K strakes. " The very plating seam that was reinforced just 2 months before Titanic's maiden voyage was the same one that Barrett observed opening to the sea!?! I felt in my gut that this could not be a coincidence. There was something about that seam that caused it to fail during the collision and I hoped to find in our experiment what that could be.
Attached to one of Carruthers's memos was a hand-drawn diagram of the modified seam by Wilding. With that diagram in hand, Bill and I were able to design three test specimens: one in the as-built 1911 configuration (referred to as the "Olympic" configuration) with steel rivets; and the second and third in the modified 1912 configuration ("Titanic" configuration) with wrought-iron rivets. Carruthers's memo made it clear that in the process of installing the reinforcing strap, all the rivets had to be drilled out and re-installed by hand, meaning that they were all wrought-iron. An extra row of holes were cold-punched into the steel to allow for the installation of the strap, making the entire landing quadruple-riveted.
Steve fashioned the specimens in his forge. The main difference in our riveting method from the original is that a hydraulic press had to be used for both the steel and wrought-iron rivets. The process of hydraulically pressing iron makes the rivet heads weaker than if they had been driven by hand. Hand riveters will "walk around" the periphery of the head when peening the rivet, rolling over with their hammers the edges of the iron as it attempts to spread outward under the strike of the hammers driving them in. If no one rolls the edges over while the hydraulic press spreads out the iron material, the iron will tend to separate. This tendency to flay creates cracking along the periphery of the rivet head, with each crack inducing potential weakness in its strength. For purposes of my test, I could accept iron rivets weaker than Titanic's…they just couldn't be stronger.
When finished, the specimens consisted of two 1' steel plates, riveted together to form a roughly 4' x 1' section of Titanic's hull at the J/K landing (size determined by our budget and the size/weight that the test stand could accommodate), the very seam that Barrett witnessed failing. The "Olympic" specimen was treble-riveted with 3 columns of rivets; the "Titanic" specimens (both iron), quadruple-riveted with 3 columns of rivets. Looking at the "Titanic" specimens, I was struck by the spacer that was used to support the topmost row of rivets in the reinforcing strap…looking at it as an engineer, the small surface area of the spacer meant that during flexing, this piece would be the easiest to move, introducing new loading angles on the rivets. We would have to pay close attention to the spacer in the "Titanic" specimens during our tests.