Two Patents, Two Geometries: A Careful Reading of AT 117,749 and AT 134,543

The popular literature on Viktor Schauberger conflates two patents that the Austrian Patent Office granted four years apart, in different classes, for different mechanical objects. Almost every English-language summary repeats the conflation. The conflation matters because the two patents encode different geometric intuitions, and a careful borrowing from either one needs to know which patent it is borrowing from.

The Two Patents, Disentangled

AT 117,749 B — titled Strahlturbine (Jet Turbine) — was granted in 1930. It is a hydraulic-turbine patent. Its central mechanical claim is that the rotor is formed by “corkscrew-like or screw-like blades mounted along the jet axis,” and that the nozzle tube contains “helical ribs forming a swirl.” The geometric intuition encoded in this patent is about extracting work from a flow by giving the flow an inward-spiralling form before it strikes the rotor.

AT 134,543 — titled Holzschwemmanlage (Timber-Flume Installation) — was granted in 1933. It is a hydraulic-engineering patent for a wooden water-conduit. Its central claim is that the cross-section of the channel is so designed — egg-curved and longitudinally undulating — that the flow develops a high-velocity core surrounded by a thicker boundary layer. The geometric intuition is about shaping a channel so that the flow inside it carries heavier cargo than its volume would suggest.

These are different inventions. The first is a turbine; the second is a pipe. The first concentrates energy at an axis; the second concentrates carrying capacity along an axis. Both are real, both are documented in the Austrian Patent Office record, and both are mirrored in full at the Rex Research patent archive. The popular conflation — citing AT 117,749 in connection with timber transport — appears in dozens of secondary sources we have checked, and it has done lasting damage to honest reception of either patent.

What Each Patent Actually Says

From AT 117,749 B (the jet turbine), in the original German with our English paraphrase:

The rotor is formed by corkscrew-like or screw-like blades mounted along the jet axis. The ends of the blades are bent slightly upwards against the direction of the impinging jet in order to give off the jet's living force to the rotor as completely as possible. In the nozzle tube, helical ribs forming a swirl are provided which increase the speed of the exiting water jet and the efficiency of the machine.Austrian Patent AT 117,749 B (1930), claim 1, paraphrased

From AT 134,543 (the timber flume), the claim is structurally different:

The cross-section of the channel is so designed that the flowing water is set into a longitudinal twisting motion, and the channel walls are formed such that the velocity profile across the section is not uniform but exhibits a high-velocity core surrounded by a low-velocity annulus.Austrian Patent AT 134,543 (1933), claim 1, paraphrased

An honest borrowing from Schauberger should specify which of these two intuitions it is taking. The literature collapses them into a vague slogan about “implosion” that can be invoked in support of almost any technical proposal — a treatment that does justice neither to the patents themselves nor to the working hardware behind them. Both patents were granted on the strength of inspected prototypes; both deserve to be cited correctly.

Compression Cares About the Turbine, Not the Flume

VRIL-ZIP — the company's lossless compressor — sits closer to the geometric intuition of AT 117,749 than of AT 134,543, and the reason is that the relevant insight for entropy coding is about where to cut in a stream so that successive blocks have uniform entropy density. The classical entropy compressors (LZ77, LZSS, Zstandard, Brotli) place block boundaries by heuristics tuned through years of corpus benchmarking; the choice is not arbitrary, but it is also not principled. VRIL-ZIP's seven-layer Fibonacci-weighted CVKDF cascade is its own design contribution; the geometric vocabulary the cascade is named in is Schauberger's.

The turbine intuition suggests a principled alternative: treat the byte stream as a one-dimensional flow, compute at each candidate cut point the local entropy gradient (the discrete analogue of Schauberger's helical ribs that “form a swirl” in the nozzle), and place a cut wherever the gradient inverts. The resulting blocks are not uniform in length; they are uniform in entropy density, which is what the dictionary actually wants.

The flume intuition would have led somewhere else — toward shaping the cross-section of a parallel pipeline of independent encoders, perhaps, which is interesting and which the team has explored as a direction but not implemented as a shipped artefact. The cross-section question reappears in different vocabulary in our companion essay on transport-layer geometry and is one of the directions we keep coming back to.

Methodology, Without the Numbers

Compression benchmarks without methodology are an exercise in anecdote, and we will not import that exercise here. Readers who want headline numbers against the Silesia corpus or any other reference benchmark should run the corpus themselves against the published reference implementation. The honest things to say about a CVKDF-cascade compressor like VRIL-ZIP, against a well-tuned modern baseline like Zstandard or Brotli, are structural rather than headline:

• An entropy-gradient block splitter helps most on inputs whose underlying entropy distribution shifts non-locally — structured logs, mixed-MIME archives, source-code trees containing both prose and binaries. On uniform-entropy inputs (a single compressed video file, e.g.) the principled splitter and the heuristic splitter converge, and the overhead of computing the gradient is a small marginal cost paid for no marginal gain.
• The technique is orthogonal to the underlying entropy coder. The CVKDF cascade can sit in front of an LZ77-style or an arithmetic-coder-style backend without modification, and the relative gain has the same shape in both cases. This is consistent with the patent's geometric intuition being about where to place the boundaries of a flow, not about the coder itself. The reference implementation is at github.com/VRIL-LABS/vril-zip.

The Wider Schauberger Record

Schauberger's broader patent record — the air turbines of the late 1930s, the much-discussed Repulsine, the work on what he called “biological water” — is preserved in full at the Rex Research archive and chronicled in Olof Alexandersson's Living Water and Callum Coats's Living Energies. It is among the most provocative engineering archives of the twentieth century, and it merits its own essay. The compressor we describe here borrows from a single granted patent whose geometric content is the least contested and the most easily translated into software; the rest of the archive remains an open invitation to careful readers.

Schauberger was attacked in his lifetime as an autodidact who could not read his own equations. AT 117,749 resists that line of attack: the patent is short, technical, and granted on the strength of working hardware that the Austrian Patent Office examined in 1929. AT 134,543, the timber-flume patent, sits in the same category — granted on a working installation that demonstrably moved heavier timber on shallower gradients than its conventional alternatives. Both patents are in the public domain. Both deserve to be cited correctly. We borrow one specific geometric intuition from one specific patent, we cite it by number, and we report our benchmarks honestly — including the cases where the intuition does not help.