If you haven’t been following the work of Knife Steel Nerds aka Larrin Thomas (related to damascus maker Devin Thomas) you are missing out. He and Pete from Cedric and Ada are my go to sources for information about steel, chemical compositions, and comparative analysis. This week, however, Larrin dropped a megaton bomb of information on steel. The article can be found here.

This article represents a sea change in publicly available information on steel. In particular, we now have scientifically valid, independently sourced CATRA data. Here is an old article I wrote on the CATRA test. Blade did something with CATRA testing a while ago and Cliff Stamp has been piecing together data for a while, but no one has had access to a CATRA machine. Larrin got access to one (actually he got a machine to keep!), he coordinated tests for a heaping helping of steel, and took steps minimize differences in geometry and heat treat. He also accounted the “coating problem” whereby a blade’s coating artificial inflated the CATRA numbers.

This data is tremendously important. While the CATRA machine has been in existence for years, it has been used mostly to do in-house testing for knife companies and the results have been closely-guarded secrets. Now, they are all in the open and we can compare steels in terms of edge retention in a meaningful sense. This is basically the Rosetta Stone for the knife world. Thanks to Larrin and go become a Patreon if you can.

Observations (its the chart about 2/3rds of the way down that has all the data)

D2: Holy Moley, D2 did quite well. Did you expect it to do better than VG10, 3V, A2, CruForge V, and Super Blue? D2 is closer to M4 than it is to O1? Clearly, if you are on budget, these data suggest that D2 is the best choice. No truly inexpensive steel is even close.

Super Blue: Yikes! Super Blue crapped to bed on this one. This exotic, import-restricted Japanese steel has a vaunted reputation for edge holding and rusts as fast a redhead gets a sunburn. Given all the hassle you’d think it would be much, much better, but it is clearly a steel in the bottom third of edge retention. Seeing as that was its calling card, there is really no reason to get a knife in this steel. V-Toku 2 falls into this category as well.

S45VN: This looks to be a very good all around steel has it has serious performance improvements over S35VN (which trades off edge retention for superior corrosion resistance) but retains the edge retention of S30V.

AEB-L and S30V: These steels show some real performance improvements with the correct heat treat and hardening. Both spike quite precipitously and that is interesting. This allows makers to buy relatively common steels and tweak the performance for different purposes. It basically makes these steels whatever someone wants them to be.

10V: like the two steels above, it has a performance spike based on heat treat, but the good thing is that even at a low hardness, it is still quite good. If you want max edge retention, this looks like a good option, but we knew that already.

ZDP-189: Its a bit old now. Its hard to get. But it is still exceptional. S90V, S110V, and S125V are its only competitors in terms of stainless steels. Its better than M390 (here represented by its chemical twin 204P), 3V, M4, S35VN, and other more common steels. I am surprised at the difference between ZDP-189 and SG2.

A2 and O1: These steels are common fixed blade steels and often mentioned in the same breath as D2, but they are clearly not on the same level of performance. I wasn’t surprised as much by O1, but I had long believed that A2 was basically an old fashioned M4. Its not. Not even close.

Vanadais 8: Yep, I need to get some of this. I don’t care what the knife is, I just need some.

Overall, I would like to see XHP and 1.4116 added as these are relatively common steels that didn’t show up. I’d also like to see if heat treat variants matter, like a Bos-heat treated 420HC, which Pete’s data has shown to be better. And, of course, I’d love to throw INFI on the machine. The most controversial steel of all time, certainly should be scientifically evaluated. Thanks a ton to Larrin and be prepared to lose an hour or two. These data are amazing.