No phantom disc.
Any local law g=f(gN) puts extra pull in the Galactic plane. The measured vertical potential wants a rounded halo instead. Local laws miss it by χ² 1000–3000 against 65–125 for a halo.
An open investigation into gravity
The vertical pull near the Sun rules out every local modified-gravity law. What survives is a response to the mass a galaxy encloses, plus a memory set at formation.
Where we stand
One hard empirical result, one coherent picture, and one decisive test still ahead. The Milky Way's vertical potential excludes any law that amplifies the local field, by a factor of 10 to 40 in χ². A response tied to the enclosed spherical mass, with a formation memory, passes the galactic and Solar-System data. Its microscopic reading is a soft glass of the de Sitter vacuum. This is a real constraint and a consistent model, not a confirmed discovery: no risky prediction has yet been verified, and several failures remain.
If you have one minute
Galaxies spin too fast for the matter we see. The usual answer is invisible matter; another is that gravity itself changes at very small accelerations. This site tests the second idea against measurements and finds it can only work in one specific form: the extra pull must follow the whole mass a galaxy encloses, never the local field, and it must switch off around anything that was born dense, like the Sun. That form passes every test we could run, fails on the faintest dwarf galaxies, and will face its decisive test in December 2026, when Gaia measures very wide pairs of stars. Nothing here is a discovery; it is a hypothesis with its numbers and its obituary conditions on the table. Unfamiliar words are in the glossary.
For the specialist, in one line
gobs = gbar + (ν(gsph/a₀) − 1) gsph with gsph = GMb(<r)/r², McGaugh ν, a₀ = cH₀/2π unfitted, applied only to systems that collapsed at g < a₀: SPARC 0.138 dex, Milky-Way vertical potential χ² 207 vs 125 (halo) and 2983 (QUMOND), Q₂ = 2.5×10−31 s−2, wide binaries Newtonian at all separations; inherits MOND's failures on ultra-faint dwarfs, clusters and the CMB; relativistic completion borrowed from AeST; microphysics a soft glass, exploratory.
The argument, in three steps
Each step is a measurement or a calculation you can inspect, not an assumption. The first is the one that removes most of the field.
Any local law g=f(gN) puts extra pull in the Galactic plane. The measured vertical potential wants a rounded halo instead. Local laws miss it by χ² 1000–3000 against 65–125 for a halo.
Let the extra pull follow the spherical field of the mass a system encloses, not the local field. No disc, no effect around a single star, and the galactic rotation curves are kept.
A system born dense (a star cluster, the Solar System) locks its medium and stays Newtonian; one born diffuse (a galaxy) responds. One rule sorts Pal 14 from the dwarf galaxies.
A picture you can hold
The medium is a glassy solid everywhere. In a galaxy most of its elements still have slack and the medium responds; where a star was born, the medium that collapsed with the natal core was strained past its slack and stays fully engaged. The Sun sits inside such a pocket, which is why the Solar System is Newtonian.
Two ways in
The same pages serve two readers. Each page opens with a plain-words box and ends with what would break it.
Start with The vacuum for the idea, then First black holes for what it would change in the early Universe, then Observations for the test that decides. Skip the equations; the boxes and the tables carry the story.
Go to Results for the vertical-potential no-go, The model for the law and the memory rule, Methods & data for the Hamiltonian, then Black holes for the limits: entropy, glass physics, the LIGO bounds and general relativity reread.
Status
Three grades, used consistently across the site.