The Missing Footnote
How an inventor disappears from the technical literature
Anyone writing today about the manufacture of polygon profiles cites Robert Musyl from 1955, with page numbers. The process by which, according to the specialist literature, these profiles are “currently mostly” made is sourced to a Bulgarian journal from 2005. The origin of the process has been published in German trade journals since 1986, with the inventor’s name.
Disclosure: one of the two authors of this text, Hans Ley, is the inventor of the process discussed here. Everything that follows rests on published sources that anyone can check.
I. The Rolling Circles
Polygon profiles connect shafts and hubs without a key. The shaft is not round but slightly triangular or square, and the hub has the same shape inside. This transmits high torques, centres itself and spares the material. The problem was always manufacture.
In the 1930s the Viennese firm Ernst Krause & Co. was the first to produce such profiles industrially, the so-called K-profile. It arises like the path of a point on a wheel rolling on another wheel. In 1955 the Viennese engineer Robert Musyl showed how today’s polygon curve, later standardised in DIN 32711 and DIN 32712, can be developed from it. His decisive step was to imagine the rolling circle as infinitely large. The tool then describes an ellipse relative to the workpiece. For the machine this meant that part of the motion had to go back and forth, with rockers, slotted links and eccentrics. Reciprocating masses limit both speed and accuracy.
Coupled rotary motions existed as well. Machines such as the “Tornomat” by Hahn & Kolb or the “Profilator” by WERA let workpiece and tool rotate together, at a fixed ratio to the number of corners. Squares and similar shapes could be made quickly this way, but only on the outside and only approximately: the faces were sections of ellipses, not exact polygon profiles. That is how a 1991 trade article on the machine discussed here describes it.
II. Two Steps
On 19 June 1982 Hans Ley and Manfred Schmidt filed a German patent application for a process (DE 32 22 991). The tool runs on a small circular path at the side of the rotating workpiece, guided either translationally or via a rotary axis with length compensation. Its angular velocity is varied periodically during each revolution by a gear train, and a second gear train compensates the resulting inertial forces. That was a big step away from Musyl’s rockers and slotted links, but not yet a motion made of pure, uniform rotations.
Ley had the idea that changed this in 1983, on his own: the three-spindle process. Now the tool no longer runs at the side of the workpiece but orbits its axis of rotation. It rotates about an axis that itself orbits a second axis at a fixed distance, and the workpiece rotates about a third. The number of corners of the profile no longer follows from the ratio of the speeds but from their difference. For a triangle the tool makes three revolutions while the workpiece turns twice. The superposition produces the same curve that Musyl had generated with rolling circles and ellipses, and both outer and inner profiles can be made. So that the new system could also modulate the rotational speed, Ley took over parts of the 1982 invention together with its priority. This became the European application of 18 June 1983, granted in 1988 as EP 0 097 346.
What is today called the two-spindle process is the reduced form of this 1983 invention: tool spindle and workpiece spindle, offset from each other by the eccentricity, both at constant speed. It was in this reduced form that Ley first demonstrated the process, at the EMO 1985 in Hanover. The three-spindle machine was finished only weeks after the fair. Heckler & Koch showed the complete machine with three spindles and a facing head at the METAV 1990 in Düsseldorf.
The idea behind it can be described without formulas. For the shape produced on the workpiece, all that counts is how tool and workpiece move relative to each other. Whether the tool turns around the workpiece or the workpiece turns beneath the tool makes no difference to the result. Whoever chooses the reference point cleverly can replace all reciprocating motion with rotation.
III. The Trail in the Literature
The process was published from the start, and where the profession reads.
In 1986 the journal “Maschinenmarkt” carried the article “Um die Ecke drehen” (“Turning round the corner”) by Paul-Gerhard Althaus and Hans Ley, on manufacturing non-circular profiles and polygons economically and to high quality. In 1987 an expanded version followed in the “Jahrbuch für Optik und Feinmechanik” under the title “Herstellung von Unrundprofilen – ein neues Verfahren” (“Manufacturing non-circular profiles – a new process”). Both describe the principle with phase images of a demonstration model. The yearbook already states what the machine looks like: “This kinematics is realised with two spindles whose axes can be shifted parallel to each other.” The two-spindle process was thus published in full detail in 1987, eighteen years before the article from Bulgaria.
In 1989 Uwe Schönwandt, professor at the Gummersbach polytechnic, proved in the “Zeitschrift für wirtschaftliche Fertigung” that the process produces exactly Musyl’s standardised polygons. His first sentence reads: “This article is based on the process described in [1] for turning and grinding polygons with the aid of rotating tools.” Source [1] is a European patent of Ley’s. Schönwandt describes the tool’s path as the sum of three pointers rotating at constant speed and shows, with the addition theorems of trigonometry, that this sum of pointers yields Musyl’s curve point by point. His abstract already states the advantage: no translational motion, only masses rotating at constant angular velocities.
In 1991 Dieter Hofmann reported twice on the machine Heckler & Koch had built from it, the FDZ 100 form-turning centre. In “Werkstatt und Betrieb” the abstract itself says: “At the 1985 European Machine Tool Exhibition EMO, the kinematic functional principle of this new process was presented by its inventor, Hans Ley.” In the VDI journal the process is called the “Ley system”. It goes back “to a patent of the inventor Hans Ley, on which Heckler & Koch acquired the rights for worldwide use”. The same article notes that the machine can already produce the polygon taper of the Capto tooling system.
The inventor, his patent, the first public demonstration and the manufacturer that acquired the rights: all of it appears in journals of the Hanser publishing house and of the VDI.
IV. A New Name
In 2006 Peter Werner Gold, then head of the Institute for Machine Elements and Machine Design at RWTH Aachen, presented a “new ICTT process” in the journal “antriebstechnik” under the title “In acht Sekunden zum Polygon” (“To a polygon in eight seconds”). The company Iprotec, he wrote, had developed it “in many years of development work”. What is described is a non-circular turning attachment with a tool spindle “whose axis is offset from the main spindle by the eccentricity e”, and what it produces are hypotrochoids. That is the two-spindle process, the reduced form of the 1983 invention. As the source for the process the article names a brochure from Iprotec. Ley, Schönwandt and Heckler & Koch do not appear. Musyl, by contrast, is expressly credited as the origin of the polygon connection.
Yet Gold knew the process at first hand. In 1985 he was one of the managing directors of the firm Reintjes in Hameln, which manufactured the parts for the very non-circular turning unit with which Ley presented his process at the EMO. Later he and Ley jointly approached Ford. How the patents then went to Iprotec and how a dissertation came out of Gold’s institute naming that company as the developer of the process, we described in Legal Expropriation.
V. The 2026 Proceedings
In November 2026 the 11th VDI conference “Shafts and Shaft-Hub Connections” takes place in Munich. The proceedings, VDI-Berichte 2477, contain two contributions by Masoud Ziaei, professor at the West Saxon University of Applied Sciences in Zwickau. Ziaei has worked on polygonal connections for many years and took part in standardising the H-profiles of DIN 3689.
His overview of the history of polygon profiles is carefully sourced. The Krause company is cited, the grinding machine of 1938, Musyl’s article of 1955 with page numbers. Then it says that H-profiles are “currently mostly” made by the two-spindle process. The source given for this is a single article: J. T. Maximov and H. Hristov, “Machining of Hypocycloidal Surfaces by Adding Rotations Around Parallel Axes”, Trakia Journal of Sciences, 2005. The second contribution in the same volume sources the two-spindle process to the same article.
None of the five publications from 1986 to 1991 appears in either reference list, nor do Ley’s patents. Of the 25 sources in the overview, nine come from Ziaei himself or from his university.
Why this is so, we do not know, and we do not speculate. All that can be stated is this: Schönwandt’s article builds expressly on Musyl, whom Ziaei cites so precisely, and names the source of the process in its first sentence.
VI. A Process Declared Unsuitable
The missing source would be a matter of academic care if it stood alone. But it stands next to a second claim. Both contributions declare turning unsuitable for many standardised H-profiles. With strongly concave profiles, “contact angles of 50° and more” could occur, requiring clearance angles on the tool that do not exist in practice. “Many standardised profiles therefore cannot be produced by turning.” As a way out, the second contribution presents a generating process for which Ziaei’s university holds a patent (DE 10 2019 000 654 B4).
We calculated, for all 18 profile variants of DIN 3689-1 with the values from the standard’s tables, how large this angle becomes at most in the two-spindle process. For eleven variants it is at most 19.5 degrees, for four variants 21 to 24 degrees, for three variants 30 to 33 degrees. No standardised profile comes anywhere near 50 degrees. Even with the less favourable oscillating turning process the largest value is 35.4 degrees. By the upper limit Ziaei himself gives, 28 to 34 degrees, 15 of the 18 variants can be turned without difficulty in both processes. What follows from his own figures is not “many standardised profiles” but three at most.
For the generating process the contribution shows no manufactured part and no measurement. Its abstract promises manufacture “without manufacturing deviations”. In generating processes, however, every deviation of the tool, whether from its manufacture, from wear or from regrinding, is reproduced directly in the profile. The contribution says nothing about this.
The result is a picture nobody in the proceedings corrects: a process whose origin goes unnamed is presented as limited, and a process for which no part has yet been made is presented as the solution.
VII. How Inventors Disappear
The pattern is not new, and it does not affect only one inventor. Whoever is cited exists in science. Whoever is not cited does not exist, even if his invention has been running in machines for decades. Between 1986 and 1991 five publications in German trade journals named the inventor. In 2006 the process was called ICTT and came from a company; in 2026 it comes from a Bulgarian journal of 2005. The name did not drop out all at once but step by step, and none of those steps had to be justified.
For the readers of a conference volume this gap is invisible. They read a careful overview with 25 sources and have no reason to suppose that one is missing. That is exactly why the missing footnote is not a slip in the margin but a statement about the history of the field. Whoever compiles the literature on a process decides to whom it belongs.
Further reading: Legal Expropriation · The General Measure · How Inventors Disappear · Who Finances the Inventors? · The Rubber Yardstick