Alien Math · a second look
A shipping lane that starts in the wrong sea
The printed text of a United States federal rule starts the Nantucket to Ambrose shipping lane in the Gulf of Mexico. A separate scan flagged that and a handful of other slipped digits. This page is the second look: five of the claims hold up against the published text or code, and the rest could not be reached.
Printed coordinates
Three slipped digits
The lane that borrows a corner from Louisiana
33 CFR 166.500(b)(1)(ii), the Nantucket to Ambrose Safety Fairway, lists nine corners. Corners 2 to 9 all sit near 40° N, between 70° and 73° W. Corner 1 reads 28°54'33" N, 89°26'07" W. That exact point is corner 1 of the Southwest Pass fairways at the mouth of the Mississippi, in 33 CFR 166.200 paragraph (28). The companion lane running the other way lists eight corners, so the ninth row looks pasted in. The section dates to 1987 (52 FR 33589).
A longitude with 122 minutes
33 CFR 165.1192(a)(1) draws a security zone at San Francisco airport. A degree holds sixty minutes. Vertex 2 has a hundred and twenty-two.
Printed this way since the rule appeared in 2004 (69 FR 34282).
A beach mouse four hundred kilometres west
50 CFR 17.95(a) lists the corners of critical habitat for the Alabama beach mouse as map eastings. In Unit 1, at Fort Morgan, one easting lost its leading 4.
A dropped leading 4 moves the corner about 400 km west. The scan claimed a second such corner in this unit; the text fetched for this check stopped before reaching it.
These are slips in the published text. They say nothing about boundaries anyone enforces: nautical charts are drawn from source data, and nobody redraws them by re-reading the regulation. For the habitat rules, which of text and map is the legal boundary for a 2008 or 2009 designation still needs checking.
Side finds
Two thermometers that read wrong
The same scan flagged two temperature-sensor drivers in open-source projects. Both hold up against the current upstream code, one with a correction.
Klipper, MAX31855 and MAX31856 thermocouples below zero
The decode in klippy/extras/spi_temperature.py turns a negative reading inside out. The raw value for −0.25 °C comes back as −2048.0 °C, and the raw value for −2048 comes back as −0.25. Correct two's-complement is adc - 0x4000. The MAX31856 path has the same pattern.
The correction: the scan blamed this decode for the "Thermocouple reader fault" shutdown, and that part is wrong. The firmware in src/thermocouple.c compares the raw unsigned value against its limits before Python ever sees it, so any reading below zero trips the range check on its own. A patch to the decode alone would still shut the printer down below 0 °C. A search of the project's issues found no earlier report of the sign bug.
Zephyr, MAX31865 platinum sensor
The driver uses the full Callendar-Van Dusen formula only when built against one C library, newlib. Zephyr's default is another, picolibc, so the default build takes a straight-line shortcut.
| True temperature | Default build reads | Off by |
|---|---|---|
| 100 °C | 98.5 °C | 1.5 |
| 400 °C | 376.4 °C | 23.6 |
| 850 °C | 743.2 °C | 106.8 |
picolibc has a square root, so the guard that hides the full formula looks stale.
The whole list
Every claim and where it stands
| Claim | Standing | Seen in the text |
|---|---|---|
| Nantucket to Ambrose lane starts in the Gulf of Mexico | confirmed | 33 CFR 166.500(b)(1)(ii), corner 1. The scan's dating, "in print by 2000", was left unchecked. |
| San Francisco airport zone, longitude 122°122'18" W since 2004 | confirmed | 33 CFR 165.1192(a)(1), vertex 2. The comparison with the draft rule was left unchecked. |
| Alabama beach mouse, two corners about 400 km off | one confirmed, one not reached | 50 CFR 17.95(a), Unit 1, easting 01503.87. |
| Klipper thermocouple sign decode | confirmed, with a correction | The decode is wrong; the shutdown has a different cause. |
| Zephyr MAX31865 straight-line formula | confirmed | The shortcut is the default path under picolibc. |
| Big Island picture-wing fly habitat, every fifth corner lost a leading 8 | untested | 73 FR 73794 (2008). Every mirror within reach cuts off before the coordinate lists. |
| Reticulated flatwoods salamander, fifteen corners lost a leading 3 | untested | 74 FR 6700 (2009), 74 pages. Same cut-off. |
| Piping plover in Texas and North Dakota, Braun's rockcress, Kauai Unit 4 | untested | Same access problem. Four claims. |
Limits
Limits of this check
- The text was read from Cornell's Legal Information Institute, which mirrors the government's electronic code. The government sites themselves were out of reach from the lab's machines.
- Searches for the exact strings turned up only mirrors that repeat the errors. That is weak support for "nobody has reported this": a search cannot see an agency's correction queue.
- The method behind the scan, looking for a corner far from its neighbours that one digit would repair, will also fire on edges that are meant to be long: offshore lanes and habitat units that hop between islands. The Kauai claim sits in exactly that zone.
- Nothing here has been sent to the Coast Guard, the Fish and Wildlife Service, or either software project.
Checked 2026-11-08. Back to Alien Math.