I’ve been working really hard on an extensive blade flex testing study, so a week ago I decided I would take a night off from doing blade flex stuff. And relax a little. I was involved in a little bit of banter about the RDL glosses, and the topic of cutting instructions came up. Which made me curious about just how often we’re actually told to hit with the edge, and how often we’re just told to hit. Because if the edge isn’t mentioned it probably isn’t something that stands out to you. Likewise if an instruction says “go out and hit with the long edge” in a pretty obvious place you likely won’t pay it much heed. So chances are the top-of-our-head recollections aren’t so reliable.
And thus I popped over to Wiktenauer and opened up the Lew gloss, and used the super rigorous Ctrl+F method to see what I found. After doing a quick search for the terms “long”, “short”, “hew”, “strike” I got the following totals and popped the below chart in the chat.
I added the 0 counts of “flat” because I was feeling cheeky. But, as the conversation kept rolling, someone brought up “what if we made an interpretation where we considered every unspecified flat”. Which sounded interesting. Certain things I’d assumed about the Lew gloss jumped out at me. Neither the zornhau nor the right zwerhau had edges specified.
But, if I was going to do this for real I wanted to do a much better survey than Ctrl+F, and picked up Cheney’s “RDL” translation book to do a cross comparison of the R, D, and L glosses. Which basically took up the whole rest of the night, and generated the first conclusion:
I’m really, REALLY, bad at taking breaks from science projects.
RDL Side By Side
So now that I was going to put in the effort to burn through the full gloss line-by-line I should record things a little better. Something like this is all pretty subjective, so I needed some sort of methodology to follow. The rules I set out before I started counting were:
- I only recorded explicit actions, not starting conditions. “Bind on him with the long edge” and “Bind on him strongly” are explicit instructions, “when you bind against” is a description of the initial conditions to facilitate the coming instructions.
- When you are instructed to attack someone (cut/strike/hew) I listed it as a cut. Anything else goes in the “other” column. The vast majority of these other actions are binding actions on the blade – winden for example – or slices.
- If something is an explicitly named technique, and the technique has an edge defined when it is explained, it counts as the edge wherever it is used. So every time it says to use a schielhau I count that as using the short edge, because when the gloss explains what the schielhau is there is explicit edge instruction.
- It only counts as an instruction if it is specific on the action. “And you can thrust or cut them” isn’t a specific action, it’s just telling you that they are now open to be attacked.
- Blade actions count for the “Other” category if they are describing exerting force with the sword. So “wind the short edge against their blade” or “press strongly against them” is an instruction to exert force. “Wind to the right” does not specify that you are pushing against a blade and wasn’t counted.
I planned to count across 3 categories:
- Cut or Other Action
- Edge Specified
And, after an hour and a half I had something.
One very important note: Doing anything like this is going to be very subjective, even if you define your rules ahead of time like I did. I’m very sure if I did this again today I would probably arrive at some slightly different numbers.
The Five Hews
It’s interesting how different glosses are explicit about different things. Here is a mapping of what is specified in each gloss. Starting on Ringeck I was prepared for a lot of explicit definition on edges, as that is the only of the three that defines the edge for the zornhau. But as you can see we only get instructions to use the long edge on a shant 4 cuts total! The glosses are not identical on which cuts are defined with an edge.
Danzig and Lew don’t actually specify what edge to use when performing the zornhau, but zorn is specified as long in Ringeck. And none of them specify what edge to hit with the taking-off. (Meyer says short!)
Though Ringeck is explicit about the long edge in the zornhau section, the tables are turned in the krumphau plays. The edge is not specified in Ringeck, but both Danzig and Lew say long. And this time they also agree on the follow up play, but this time it’s an agreement that it should be with the short edge.
But the real one that caught me by surprise was the uniform lack of specification on the left zwerhau, which is universally associated by the community as a long edge strike. Is this omission common to all glosses? Let’s phone a friend, and see what 3227a has to say:
“ And it besets like a crossbar to both sides: with both edges, the back and the front…”
“…only that for someone in the thwart-hew, the flats of the sword turn: one above or upward, the other below or downward; and the edges to the sides.”
“…he shall then immediately win the after-strike in a rush directly without pause with the thwart-hew to the other side with the forward-edge…”
Ok, so 3227 definitely doesn’t leave anything to doubt that it wants the left zwer to be done with the long edge. Which makes it all the more weird that it’s not mentioned in R, D, or L.
The n-5 cuts
Outside of a whole bunch of right zwerhau, what were the most common types of cut that don’t have an edge? And of course I’m kicking myself for not keeping more detailed records. But, as always, you don’t really know what the important categories to keep track of are until you’ve been through the exercise once.
Just going from my recollection outside of the 5 Hews the most common place to see a non-specified edge is when it’s describing a fencing principle, rather than instructions on specific sword technique. For instance, in the description of the cut form of nachrisen you’re just supposed to hew after them in the generic sense, no target or edge is given. Or you see things like this:
“This is when he mis-hews in front of you, pursue him with a hew to the upper opening. If he then rises and binds on you below at the sword, not as soon as one sword clashes onto the other, fall from the sword with the long edge over his arm, and press him from you like that..”– Ringeck, Cheney Translation, p115
In the first bolded segment we see a cut with a non-specified edge, which is describing an action we’ve already seen earlier. It feels kind of like the setup to the play, rather than explicit instructions on how to cut, but it gets a tally according to the rules I laid out before I started. The second bolded statement is much more indicative of what happens when we get into the meat of the instructions. There is a new sword action described – falling on to the arms when they bind on you firmly from below – and we get a corresponding instruction on which edge to use.
So, what makes up the “Other” bin, the one encompassing everything isn’t a strike? I would estimate that the bulk of it is either slices or instructions on how to bind against the blade to do things. Interestingly enough the slice isn’t mentioned for edge in a lot of places, but it follows a similar pattern to edge instructions. In multiple places we are instructed “if they do this, then slice them”. When the slices are taught in their own section, the slicing off and hand pressing parts, they are diligent about describing the edge.
Likewise I was surprised with the amount of times it said “bind with the long edge”, where I didn’t remember an edge was specified. I didn’t spend a lot of time thinking about these as I went, because it was cuts I was after, but I did recall being surprised at how often the slice didn’t mention an edge, and how often pushing against their sword did.
So, what’s my conclusion after all this? The RDL glosses are way more prescriptive about using the edge of the sword than I would have guessed. Just about every single case of an unspecified edge in a cut is a result of:
- It is one of the 5 Hews that has an edge specified in another gloss.
- It is part of an illustration of a fencing concept or setup, rather than instructions on how to do something with your sword.
- It’s a left zwerhau.
Now, one could argue that being a purist in Lew you could be doing your zorn with the flat. BUT, then you have to consider that each gloss leaves out multiple pieces of information found in the others. If you claim that a flat zornhau is valid, because it’s not defined in Lew, then I can claim that an unterhau is a zornhau because the descending aspect isn’t specified in Ringeck. 😂 (Which fits into something I’ve wanted someone to do for a while: create the wackiest possible full gloss interpretation by using the section from each that specifies the least amount of information.)
Likewise the “Other” section wasn’t something that I kept stats on in detail, but it was really interesting to see things like slices showing up a decent chunk without edges defined. So obviously at this point I realize that I’ve wasted enough hours on something that was just supposed to be a throw away joke. And I call it a fun rabbit hole, put it down, and go do something else.
Even More Counting
Haha, calling it good would be what I would do if I was smart and not empirically obsessed. So now I want to go through again and count (we switch from past to present tense here, because I started writing the article between the first RDL count and what I’m about to describe next.)
Going through I’ll be using the same rules as before, but keeping much more detailed information. Mainly if an unspecified technique is specified in another source, and the nature of the unspecified actions. I’ll also be adding good ol’ 3227a into the mix.
The chart differentiates a cut that is defined as being performed by the edge in a different source. So now all three RDL glosses get to benefit from the 3227a’s ultra-explicit fascination with the edges of the zwerhau. And Ringeck gets to take credit for almost every long edge device that it omits the details on.
Unspecified – Example
We now have a much higher resolution on the “unspecified” column from earlier. Being able to borrow instructions helped a bunch with binning the left zwerhau. We can also see that the majority of the unspecified attacks go into what I call the “examples” bin. This would be instructions such as:
Another pursuit: When he hews to you from above, let his sword go down to the ground with the hew, then pursue him with a hew in above to the head sooner than he rises with the sword.– Ringeck, Cheney Translation p107
Where it doesn’t really sound like we are being instructed on how to use a cut other than “they are open”. Sections like:
This is when you approach him on guard, if he then hews to you below to the lower opening, do not parry him, but rather hew in above strong to his head.
If he then rises and binds to you below at your sword, remain strong with the long edge on the sword, and work nimbly to the nearest opening, or let him work, and you come indes, so you land a hit on him.– Danzig, Cheney Translation p119
The first part does not specify how to cut to the head. But the method of cutting is not the point of that section, the instruction on uberlauffen is. And we see in the follow up section an explicit instruction on blade interactions shows up, and the edge to use becomes prescriptive again. Which is a similar pattern to these sections I have classified as Unspecified – Example. (It also strongly suggests the cut was with the long edge, but that is far too big a leap for the counting rules I established beforehand.)
Unspecified – Direct
Unspecified – Direct, on the other hand, are those where we clearly have an example of bladework but no edge is defined. None of the glosses mention the edge that taking off above should use. Possibly because it could be long or short? Meyer describes it as short. While I don’t think shoehorning 16th century Meyer into the 15th century gloss-times is good practice, it does show that it isn’t outside the realm of posibility.
Other than that they seem to be weird omissions. The following appears in Danzig only, and it’s in the middle of a group of 6 plays that all specify edges.
When you have bound a fencer to his sword, if he then strikes around from the sword with the lateral to the other side, fall to him with the long edge into his hand or onto the arms, and press his arms from you with the sword with the slice, and strike him out of the slice from his arm with the sword onto the head.– Danzig, Cheney Translation p56
We get explicit instruction to push the arms away, and then are told to just “hit them in the head”. Which seems inconspicuous enough, but it’s actually a weird oddity. In just about all cases something like this would specify which edge to strike with.
Another passage for comparison from Lew:
When he has parried you and does not want to withdraw himself from the sword, and he means to allow you to come to no technique, do as if you would like to withdraw yourself from the sword, and retract your sword to yourself until half the blade, and quickly rise with the sword and hew quickly with the short edge or with duplication to his head.
Another: When he has parried you, wrench up with your sword at his sword’s blade upward against his point as if you would like to take off above, then remain at the sword, and hew in straight at the blade again to his head.– Lew, Cheney Translation p100
The first part of the section is more typical of how we are instructed, the edge of the hit is specified. The second is the weird one, in which we aren’t given instructions. This could be because nowhere is it specified which side is being parried against and the edge is side dependent. It could be part of the same ball-of-wax that leads to the taking off above not having an edge, since you’re supposed to act like you are taking off above. Or it could be a plain omission and reading too much into it is a fool’s erand. ¯\_(ツ)_/¯
The first play is also difficult to count; we have an explicit instruction on the short edge cut. And we also are given the option for duplication, a named technique with edges specified. Long on one side, but short on the other. But which side does this refer to? We don’t know which side we are parrying on. So in the end I didn’t count it, despite it being an explicit instruction to an edge.
Wrapping Up (Conclusions II)
I don’t want to spend too much time looking into the void that is exact word choice in HEMA sources, because I’ve seen that is definitely a void that will end up staring back at you. Playing a mind reading game of important-or-fluke is not something I particularly enjoy, which is why I stick to science stuff with stickman memes. I just thought that the count showed something that I didn’t expect, and it would be a nice quick article to share.
Not wanting to go deeper is partially a result of me not wanting to over-analyze text. One reason for this is that a fixation on counts leads to bad conclusions on how to actually fence. Frequency of mention is NOT equal to frequency of use, something that is obvious to anyone who fences regularly.
So overall I’m not a huge fan of frequency counts as “proof” of anything. But what they can do is generate new questions for us to ask. I did some counting of targets in How Did The Sources Say We Should Weight Targets?, but the rule sets generated were just food for thought rather than something I would ever advocate for. Weighting by this metric would be dumb, but at the same time if you see behavior in a population that goes against this (such as torso cuts being a mainstay) it behooves us to notice the discrepancy and ask some more questions.
A better use of counting is to look for patterns that aren’t obvious when you are reading play by play. The counts that Keith Cotter-Riley did point to some interesting correlations between actions and hand position in Meyer’s longsword. (He was kind enough to let me publish his data: How Does Meyer *Illustrate* a Longsword). Don’t use this to form interpretations, but maybe keep it in the back of your head during your next read through*.
Wrapping Up More (Conclusions III)
But we get to the biggest reason, by far, to not read into this. The huge variability in the counts.
“One very important note: Doing anything like this is going to be very subjective, even if you define your rules ahead of time like I did. I’m very sure if I did this again today I would probably arrive at some slightly different numbers.”Sean Franklin, SwordSTEM, “Turns Out Lichty Glosses Are Oddly Specific About Edges”
I did get different numbers the second time. Because I had to make the judgment calls all over again. And interpret the rules slightly differently all over again. And I had to risk putting checkmarks in the wrong box of my overly categorized tracking sheet all over again. So while I’m pretty confident in the trends I’m seeing, the exact numbers are going to shift a bit from count.
If I wanted to do this for real I would:
- Double the number of the rules
- Do it with a group of people. To force discussion/consensus on each item, and to be more rigorous in record keeping.
- Not be doing it up until 2 am. Probably best to budget a whole afternoon (with food) given how much slower the group would make it.
But I’m pretty confident I wouldn’t be drawing any different conclusions than I already have. Because I just used the counts to point me into observing something new that was already in front of me, and I hadn’t been taking notice. (If anyone wants to go nuts and do this the super formal way, I would love to publish those results.)
What’s next? Sometime after I’ve recovered I’d like to do the same exercise for Meyer, who does have more variability due to flat strikes being included in the material. I also wasn’t able to include Meyer in the target areas article, because the longsword is all targeted to the head.
Stats For Nerds
You may have noticed I didn’t actually publish the counts for the second set of counts. This is because it’s the percentage of mentions that is important. But for the sake of transparency, here they are. I also didn’t include the non-cut actions in the article, because I’ve about exhausted my interest in talking about the subject.
In order to keep the data straight I jotted the mentions I was counting into a text editor with a simple 3 character code code. Then I could just filter into a spreadsheet to get the counts. And it’s much more traceable to go back and see what I was and wasn’t counting. (In some sections it says something like “do the doubling”, or “do the snapping”. Both of these actions do have explicit edges assigned to both left and right version. But if the text doesn’t say if I’m on the left or right to begin with I can’t count it in either the long or short tally. So it gets ignored.)
- 3 – 3227a
- R – Ringeck
- D – Danzig
- L – Lew
- C – Cut
- S – Slice. And action with the sword pushing on the body
- W – Wind. When we are instructed to wind against their blade. This does not include sections where you are told to wind to a direction relative to you without mention of a blade interaction as you are winding.
- B – Bind. Any time you are pushing against their sword. Remember I’m only counting instructions, not all the times the section starts with “when you’re binding, do <x>…”.
- L – Long
- S – Short
- * – Unspecified in this source, long/short specified in another.
- ? – Unspecified. Instruction to do an action, but edge not mentioned.
- ?e – Unspecified, example. Used as part of an example of a concept being used, where that concept is the focus of instruction.
(I’m lying. This is count #4, not count #3. Count #3 was supposed to be like #2 but better. And I wasn’t happy with the recording so I switched to this method of notation, rather than doing checkmarks on a grid with the different action types.)