I've spent the past while reading through written sources on how the five major Taijiquan traditions move their feet. Not what a stance looks like once it's finished, but how the weight gets from one leg to the other, and where the turn actually happens.
Three things came out of it that I didn't expect. One of them makes me think a lot of how we talk about styles is wrong.
There are three different kinds of "difference," and we keep mixing them up
Terminology comes first. In the Wu Jianquan line there's a whole vocabulary built around 隅步 and the pair 正弓步 / 隅弓步. Wu/Hao material doesn't use that vocabulary at all — it lists nine stance types and talks about 虚实分明 instead. Sun style talks about 开合活步, 进步必跟, 退步必撤. Same physical activity, three naming systems. That's a naming difference, and it doesn't tell you whether the feet go anywhere different.
The second kind is how the movement is actually made. Turning is the clearest case. Yang teaching material works through 展脚, 扣脚 and 拧脚, and then disagrees with itself about which part of the foot is the pivot. Meanwhile the Chen small-frame text says 以脚跟转换 and 脚踏忌八丁: parallel feet, no splayed, no pigeon-toed. Those aren't two labels for one event. Something different is happening.
The third kind is numbers — how wide, how many degrees. This is the layer that travels worst.
A term turns out to be a poor stand-in for a technique. That's the thing I kept running into.
Variation inside a style is often bigger than variation between styles
This was the surprise.
Wu style isn't one thing. There are at least three regional lines: the Beijing and Láizhōu teaching descending from Wang Maozhai, the Beijing line through Yang Yuting and Li Bingci, and Wu Jianquan's own Shanghai line. There are documented differences in how many times Cloud Hands is performed, how 退步跨虎 finishes, and how large the steps run. The 1988 45-form competition routine was assembled on purpose to draw on both the northern and southern approaches, which tells you the split was real enough to need bridging.
Chen is the same story. Large frame and small frame don't follow the same foot rules, and the small-frame text is explicit about it while the large-frame material is a separate line of teaching.
Yang is maybe the clearest case, since it's the one everybody thinks they know. The traditional routine and the 24-Form Simplified aren't "the same thing, shortened." Take closing. The traditional Yang routine is required to finish in the same place it started, and the source tells you how: adjust your spacing during the last Repulse Monkey and the last Cloud Hands. The 24-Form keeps the direction requirement and drops the position requirement. Same idea, two levels of obligation. One thing worth keeping straight: the 24-Form is a 1956 state-issued routine, not a Yang lineage transmission. Calling it "Yang style 24" blurs a distinction the sources take seriously.
And Wu/Hao: the two-axis pivot description for 实转步 that gets quoted online shows up in one documented transmission. In other material it's just a name in a stance list.
"Yang style does X" isn't a complete sentence. You need style, routine, lineage.
A lot of the numbers people quote aren't style parameters at all
The 45° rule. In Yang teaching it concerns the angle between the rear foot and the front foot, in a specific context — and an analysis of Yang Chengfu's standardized frame gives a range around 45–80°, not a fixed 45. The Sun style line about turning half a face to the left at forty-five degrees, which circulates widely, comes from a standing-post opening posture, not from a rule about Sun stances generally. Somewhere along the way 45° got promoted from a teaching context into a law of Taiji. It isn't one.
The 70/30 weight ratio. I went looking for it in Wu/Hao's classical texts. Wu Yuxiang and Li Yishe write about 虚实 quite clearly and give no numbers at all. The "six or seven tenths on this leg" phrasing is somebody describing one moment in one posture. The number isn't fake. The generalization is.
Bow stance width. Inside the 24-Form alone, sources give under 10 cm for 揽雀尾 and 闪通臂, around 30 cm for 搂膝拗步, and 10–30 cm for 野马分鬃. One routine, one stance name, several widths. That makes it a posture parameter, not a style parameter.
24-Form direction angles, for good measure: two teaching texts disagree about them. One gives the right heel kick as roughly 20° south of east, another around 30°, and they place the left kick on opposite sides. That's a text-to-text disagreement. Nothing to do with styles.
What I do with this now
When I read a description of stepping, I ask three things. Before the foot moves, which leg is carrying me, and when does the weight arrive? At the turn, does the source say heel, ball of foot, or rotation through waist and hip? And how does this step connect to the next one — follow-step, scraping cycle, feet drawn back together?
Those three questions work across all five traditions without forcing them into one set of rules.
I wrote the whole thing up with the sources kept in separate categories (classical texts, lineage publications, formal standards, secondary references, modern teaching notes) and the disagreements left as disagreements rather than averaged away. There's a free post on the blog that lays out the argument, and the full comparative study — the tables, the numerical parameters with their routines attached, the list of what's still unresolved — is in the member library.
Read the free post on the blog
If you practise one of these styles and something in your line is described differently, I'd like to hear it. The whole point of the exercise is that the unit of comparison should be narrower than "style," and the fastest way to test that is for people to say "in our line it's this instead."
