For each n two numbers are known: the side of the best packing anyone has found, and a floor that a proof says no packing can beat. Where they meet, s(n) is settled; elsewhere the truth lies in the gap, and for most n the gap has barely moved in forty years. Floors are collected for n ≤ 324; past 100 they come from general bounds and the families s(k² − c) = k.
| date | floor | by | note |
|---|
The chart draws each packing and each floor on the date it was made public, and only those that beat every earlier one; click a packing's dot to open it. The list includes the rest. Others' 2026 certificates are linked as their authors published them; Sources §5 has the details.
Click an n to open its best packing in Explore.
| n | best packing | found by | proven floor | proved by | gap | status |
|---|
Upper bounds and their dates are read from David Ellsworth's catalogue, and for packings posted since (from mid-September 2026) from the posts themselves: jlevy/squares' register and the registration requests still open there, credited to whoever posted each packing first. Every packing on this page opens in Explore. Lower bounds from Erich Friedman's survey and the papers it cites, with the 2026 computer-checked certificates (unrefereed) marked. Bounds of the form ⌈√n⌉ that follow from area alone are labelled as such. Nagamochi's 2005 theorem (the n²−1 and n²−2 families and a general bound) is listed but never used as a floor: its published proof rests on a lemma shown false in 2026 (Sources §3); its families are re-proved by Karakuş and chelokot, and Karakuş's weaker general bound is used instead.