The Repository Root Cause

Alaska Airlines 261: A Nut Built to Wear Out

November 2, 2026 · Runtime 10:45 · Watch on YouTube

On 31 January 2000, Alaska Airlines Flight 261 lost pitch control off California when the threads of one nut in its tail had worn away. The nut was designed to wear. Grease was supposed to keep the wear slow and a measurement was supposed to catch it, and both intervals had been stretched with the regulator's approval. In 1997 that measurement found this nut at the limit.

Transcript

A nut built to wear out 0:00

Alaska Airlines Flight 261 was held up by a nut built to wear out. In 1997, a check found it at the limit, and it flew two more years. We read the safety board's report page by page, to find where that wear should have been caught.

Puerto Vallarta, 31 January 2000 0:17

On January 31st, 2000, a McDonnell Douglas MD-83 left Puerto Vallarta for San Francisco and Seattle, with eighty-eight people on board. About thirteen minutes after takeoff, climbing through about twenty-three thousand four hundred feet, the horizontal stabilizer moved a fraction of a degree. Then it stopped. It would not move again for two hours and twenty minutes.

One jackscrew for the whole tail 0:41

On this airplane the whole horizontal tail moves. It sits on top of an eighteen-foot fin, hinged at the back. At the front, a single jackscrew raises and lowers it. Two electric motors turn a steel screw through a nut held in the fin. As the screw turns, it climbs or descends through the nut, and carries the front of the tail with it.

Threads designed to wear 1:02

The screw is case-hardened steel. The nut is an aluminum-bronze alloy, softer, and that is on purpose. In the words of the original fault analysis, its threads were designed to wear. A new nut thread is about fifteen hundredths of an inch thick. Boeing told the investigators it had taken the wear into account, and defined maintenance procedures to manage and inspect it.

Grease and the end play check 1:24

So the design rests on two maintenance tasks. Grease, to keep the wear slow. And a measurement, the end play check, to see how much has worn. The check pulls the tail against the screw, and a dial gauge measures how far the screw moves in the nut. A new assembly shows three to ten thousandths of an inch. The limit is forty.

Jammed, then the first dive 1:45

For two hours the crew worked the problem at cruise, and talked with company maintenance by radio. At eight minutes past four, the captain said it appeared to be jammed. A minute later the autopilot was disengaged. Within four seconds the stabilizer moved to two and a half degrees nose down, and the airplane dived for about eighty seconds, from thirty-one thousand feet to about twenty-four thousand. The pilots pulled it out.

North of Anacapa Island 2:13

For about ten minutes the airplane flew on, descending, while the crew prepared to land at Los Angeles. At nineteen minutes past four, the recorder picked up a series of thumps, then an extremely loud noise. The airplane pitched down and rolled inverted. It struck the Pacific about two point seven miles north of Anacapa Island. All eighty-eight people on board were killed.

The thread wrapped around the screw 2:40

The jackscrew was recovered from the sea floor. Wrapped around the middle of the screw was a long spiral of metal. It was the nut's thread: twelve pieces, about three quarters of its original length. Inside the nut, where the thread had been, the bore was almost smooth. The board found that about ninety percent of the thread's thickness had worn away before the rest sheared off.

How the end happened 3:06

The report reads the end this way. The worn thread was sheared off a little at a time, and the last of it went during this flight. The board could not determine exactly what caused the jam. In the first dive, the screw stayed in the nut only because a stop at its lower end caught against the nut. Ten minutes later, a titanium tube inside the screw broke in fatigue, the brackets at the top of the fin failed, and the tail moved far past anything the screw allowed.

No effective lubrication 3:36

Then the board looked for grease. On the working length of the screw, its metallurgists found none, fresh or old. Inside the nut, none either. In the nut's grease passage, a dry, claylike residue. The board concluded there was no effective lubrication on the screw and nut at the time of the accident. That is the probable cause it names: excessive wear, from insufficient lubrication by Alaska Airlines.

Twelve times the expected wear 4:05

How fast had it worn? The board expected a nut like this to lose about one thousandth of an inch for every thousand flight hours. From its last check in 1997 to the accident, this one lost about twelve. Twelve times the expected rate, on a part whose whole design depended on the rate being slow.

From 500 hours to eight months 4:27

Grease was the first defense, and its interval had been stretched. The original Douglas documents asked for lubrication every three hundred to three hundred and fifty flight hours. At Alaska, in 1987, it was every five hundred. Then every thousand, every twelve hundred, every sixteen hundred. In 1996 it became once every eight months, with no flight-hour limit at all, which at the time meant more than two thousand hours.

Reviewed and accepted 4:59

The FAA's principal inspector for the airline reviewed and accepted that change. The justification was a new industry guideline allowing fifteen months. Boeing's design engineers, the report says, were not consulted about that guideline, nor aware of it. Alaska was the only U.S. airline with a calendar limit for this task and no hours beside it. A longer interval did not wear the nut by itself. It made one missed or poor greasing far more costly.

The end play check, stretched 5:31

The second defense was the measurement. Douglas first recommended it every thirty-six hundred flight hours. Later guidance doubled that, to about seven thousand hours or thirty months. Alaska checked every twenty-six months, about sixty-four hundred hours, until 1996. Then its heavy maintenance check was extended, and this check went with it: thirty months, about nine thousand five hundred and fifty hours.

Approved with the heavy check 5:58

The board says the FAA, approving the heavy check extension, effectively approved this one too. In flight hours, it was a fifty-five percent increase. The tasks inside the heavy check were not looked at one by one. Without the extension, the board found, this airplane would have had its end play checked at least eighteen hundred to two thousand flight hours before the accident.

Oakland, September 1997: 0.040 6:22

And one check did happen. September 1997, the airline's maintenance base in Oakland. On Saturday the 27th, a mechanic and an inspector measured the end play, and wrote it on a work card. Forty thousandths of an inch, the maximum allowable. Planned action: replace the nut.

Re-evaluate test: 0.033 6:41

By Monday, parts were being ordered, and the departure looked doubtful. The engineering order cited on the card turned out to apply to another airplane model. On the Tuesday graveyard shift, the lead mechanic crossed out the replacement and wrote: re-evaluate test. The new reading was thirty-three thousandths. Rechecked five times, with the same result. The airplane left on October 2nd.

Right on the edge 7:05

Asked later, the inspector put it plainly. Forty thousandths would be right on the edge, he said, and asked how much further you have to go, with two more years before the next check. The lead mechanic said the card should not have been generated to begin with. The airplane flew another two years and four months.

A check that can be off by 0.030 7:26

The board did not decide which number was true. It found the check itself unreliable: readings can be off by as much as thirty thousandths, in either direction. That is against a limit of forty, on a part that starts at three to ten. Both 1997 readings, it says, may have been lower than the real end play.

The fixture 7:51

There was one more question, about the tool. Alaska's only fixture for this check was made in-house, with no inspection records. In tests, fixtures the airline made gave lower readings than Boeing's, by up to about five thousandths. Alaska says its fixture came from Douglas. The board says the evidence does not support that, and that it could not determine whether the fixture changed this reading.

What the backups did not cover 8:17

Which leaves the design. There is a backup inside the screw, the titanium torque tube. There are two thread spirals, and more thread than strength requires. Each protects against something. None protects against the nut's thread wearing away. The designers assumed that at least one set of threads would always be intact and engaged, and the loss of the nut's thread was not considered.

No such thing as a wear-critical item 8:44

At the public hearing, Boeing said a nut worn to the forty thousandths limit remains ten times stronger than the maximum design loads. The FAA's senior certification engineer said wear is not considered a mode of failure, and that there is really no such thing as a wear-critical item. We presume, he said, that the maintenance requirements are complied with.

Probable cause and contributing factors 9:07

The board's probable cause names the thread failure, and the missing grease. Then three contributing factors. The extended lubrication interval, and the FAA's approval of it. The extended end play interval, and the FAA's approval of it, which let the wear progress to failure without the opportunity for detection. And the absence of a fail-safe mechanism against total loss of the nut's thread.

What changed, and what did not 9:34

Eleven days after the accident, the FAA ordered lubrication every six hundred and fifty flight hours, and end play checks every two thousand. The board asked for an interval giving at least two opportunities to detect excessive wear, and for a design review to eliminate the catastrophic effects of total thread loss. In 2001, the FAA wrote that talks with Boeing had yielded no firm conclusion that a design change was warranted.

A promise about time 10:04

A part designed to wear is a promise about time. It promises that the wear will be slow, and that someone will look before it is fast. Here, both promises were stretched, one approval at a time, and each step looked small on its own page. Next case is a repair. China Airlines Flight 611, May 2002: a Boeing 747 broke apart in flight, from a fatigue crack beneath a patch put on after a tail strike in 1980. The patch did not reach past all the damage, and the inspections that followed never found the crack.

Description and sources

Alaska Airlines Flight 261, an MD-83, lost pitch control off California on 31 January 2000; all 88 aboard were killed. Its whole horizontal tail moves on one jackscrew, a steel screw turning through an aluminum-bronze nut whose threads were designed to wear. The NTSB found no effective lubrication and about 90 percent of the thread worn away. The lubrication interval had grown from 500 flight hours to eight months with no hour limit, and the end play check from about 6,400 to 9,550 hours, both approved by the FAA. In 1997 that check read 0.040 inch, the limit; it was redone at 0.033 and the airplane flew on.

PRINT-READY, FROM THIS CHANNEL

The Failure Atlas, Vol. 01 — Tacoma Narrows · Citicorp Center · Millennium Bridge · Apollo 13 · the 2003 blackout · Hyatt Regency

https://therepository.gumroad.com/l/failure-atlas

PRIMARY SOURCES

National Transportation Safety Board, Loss of Control and Impact with Pacific Ocean, Alaska Airlines Flight 261, McDonnell Douglas MD-83, N963AS, January 31, 2000, Aircraft Accident Report NTSB/AAR-02/01 (adopted 30 December 2002; public domain) -- read directly (1.1, 1.6.1-1.6.3, 1.16.2, 1.18, 2.2-2.6, 3, 4): https://www.ntsb.gov/investigations/AccidentReports/Reports/AAR0201.pdf
Next episode: China Airlines Flight 611, 25 May 2002 (Aviation Safety Council report ASC-AOR-05-02-001, to be read directly): https://www.faa.gov/sites/faa.gov/files/2022-11/CI611_English_VOL_1.pdf

ARCHIVE PHOTOGRAPHS — the real structure, with title, author, date, archive and licence. CC BY items are used under that licence, cropped and reframed for the screen.

Figure 12c — NTSB — 2002 | NTSB/AAR-02/01 | Public domain
Screwshavings2 sm — NTSB — 2008-02-27 04:54:09 | Wikimedia Commons | Public domain
McDonnell Douglas MD-83 HB-IUG Crossair — Steve Knight — 2000-10-29 | Wikimedia Commons | CC BY 2.0
Figure 3 — NTSB — 2002 | NTSB/AAR-02/01 | Public domain
Figure 5 — NTSB — 2002 | NTSB/AAR-02/01 | Public domain
Figure 8 — NTSB — 2002 | NTSB/AAR-02/01 | Public domain
Defense.gov News Photo 000202-N-5961C-005 — Petty Officer 1st Class Spike Call, U.S. Navy — Taken on 2 February 2000 | Wikimedia Commons | Public domain
Figure 12a — NTSB — 2002 | NTSB/AAR-02/01 | Public domain
Figure 13 — NTSB — 2002 | NTSB/AAR-02/01 | Public domain
Figure 16 — NTSB — 2002 | NTSB/AAR-02/01 | Public domain
Figure 20 — NTSB — 2002 | NTSB/AAR-02/01 | Public domain
Figure 15 — NTSB — 2002 | NTSB/AAR-02/01 | Public domain
p. 152, 2.4.1 — NTSB — 2002 | NTSB/AAR-02/01 | Public domain
Figure 11 — NTSB — 2002 | NTSB/AAR-02/01 | Public domain
Figure 10 — NTSB — 2002 | NTSB/AAR-02/01 | Public domain
p. 24, 1.6.2.2 — NTSB — 2002 | NTSB/AAR-02/01 | Public domain
p. 180, 3.2 Probable Cause — NTSB — 2002 | NTSB/AAR-02/01 | Public domain

The 5 technical plates in this video are illustrations generated for the channel by a diffusion image model, styled to match its cyanotype identity. They are diagrams of the system — the photographs above are the archive, the drawings are ours, not photographs of the hardware, and no person is depicted in any of them.

The narration is synthetic (Google Gemini text-to-speech), checked word by word against the script. Every diagram is drawn in code.

Root Cause investigates why engineered systems fail, using official investigation findings and primary technical literature. Sources for this episode are linked above.

#aviation #engineering #rootcause