In late February 1910, two Great Northern Railway trains were held for about a week on a siding at Wellington, Washington, near the west portal of the Cascade Tunnel. Snow and slides closed the line in both directions, and the telegraph wire that ran beside the track went down with it. In the early hours of March 1, an avalanche carried both trains into the ravine below. The toll most often given is 96 dead. This piece examines how losing the wire shaped the decisions made that week, and what that means for agencies whose backup links follow the same corridor as the hazard.

Stevens Pass in February 1910: the line, the town and the two trains

The Great Northern Railway crossed the Cascade Range at Stevens Pass. The original Cascade Tunnel opened in 1900 and carried trains under the summit. Wellington was a small railroad settlement at the west portal of that tunnel. It sat on a steep mountainside above the valley of the Tye River and existed mainly to keep the line open in a place where winter snowfall was measured in tens of feet. West of Wellington the track descended toward Scenic, a lower railroad point and hot springs resort, and continued from there down the valley toward Everett and Seattle.

The two trains at the center of the disaster were the Spokane Local, a westbound passenger train, and a fast mail train. Both were held at Wellington around February 24, 1910, after slides blocked the line farther west. Over the following days the storm continued, and more slides came down on the track on both sides of the town. The railroad fought the snow with rotary snowplows, which cut through drifts and slide debris and threw the snow clear of the track. Accounts of the week, including Gary Krist’s 2007 book The White Cascade, describe the plows losing ground as slides buried sections faster than crews could clear them, with plows and equipment trapped between blockages. Both trains ended up on a siding along the mountainside above the ravine, with passengers and crew living aboard and in the few buildings of the town while they waited for the line to open.

The setting matters for the communications story, because Wellington had only one connection to the outside world. A single railroad corridor ran through the tunnel to the east and down the mountain to the west. The telegraph wire, the trains, the plows and the people who might repair any of them all depended on it. Once slides closed it in both directions, the town was effectively an island on a mountainside, and the only way to send or receive information was the same way people could leave, on foot through deep snow and across active slide paths.

Ninety-six and the other numbers: what the death toll rests on

The figure most widely given for the Wellington avalanche is 96 deaths. HistoryLink, the online encyclopedia of Washington state history, uses 96 in its account of the disaster. On that figure, Wellington is generally described as the deadliest avalanche in United States history, and I have not seen a reputable source dispute that ranking. When I cite the toll, I give it as 96 and name the source, because that is the count with the broadest acceptance among historians of the event.

You will see other numbers. Some popular accounts and later retellings give totals above 100, and some round to “about 100” or “more than 90.” I have not been able to trace the higher figures to a published counting basis, so I treat them as variant totals and not as a competing settled count. Anyone who wants to use a different number in training should find out who published it and what it counts before putting it on a slide. A total that counts confirmed recovered bodies answers a different question from one that also counts people believed to be aboard but never found.

The record allows some caution about how the count was built, without my claiming more than it supports. Bodies came out of the ravine over a long period, by most accounts into the summer of 1910, and newspaper casualty lists in that era were assembled from whatever reached the reporters. That alone would produce scatter in contemporary totals. In my judgment, a count that includes railroad workers hired for snow fighting is harder to close than a passenger manifest, because temporary labor was not always recorded by name. I offer that as a general reason counts of this kind drift. I have not seen a historian document it as the specific cause of the variation at Wellington.

The telegraph beside the track: one path through one hazard

A railroad in 1910 ran on the telegraph. Train movements on a single-track mountain line were controlled by timetable and by written train orders. A dispatcher at the division office sent those orders in Morse code to operators at stations along the line, who copied them and handed them to train crews. The wire also carried every other kind of operating traffic: reports of slides, requests for coal and men, the location of plows, and messages between the officials on the mountain and the division headquarters below. For the people at Wellington, the wire was the link to anyone who could send help or make a decision.

The wire was strung on poles along the railroad right-of-way, and for ordinary purposes that made sense. The stations it served were on the track, linemen reached breaks by riding the track, and the railroad already owned the land. In a storm like the one in February 1910, the arrangement meant the communications path ran through exactly the same slide paths and drifts that were closing the track. A slide that buried the rails could take the poles with it, heavy wet snow and wind could bring down wire elsewhere, and a lineman sent to find the break had to travel the same blocked corridor to reach it. Accounts of the week describe the wire failing during the storm and the town losing contact for long stretches, with service coming back only intermittently when a section could be patched.

Engineers would call this a common-mode failure. The event that created the emergency also disabled the means of reporting it, and it did so for the same physical reason, because both the track and the wire sat on the same mountainside under the same snow. To my knowledge, no alternative path existed. Railroad radio did not come into routine use until decades later, and any telephone circuit through that corridor would have been exposed to the same slides as the telegraph. The railroad did not have a backup that failed. It had a single path that went down along with the thing it was meant to report on.

The modern version of this problem is common, and agencies tend not to see it until it happens. A fiber backhaul laid in the shoulder of the highway that floods, a microwave hop to a mountaintop site served by one access road, or a backup circuit leased from a carrier whose cable shares the same bridge are all versions of the Wellington wire.

Deciding on partial information: the mountain and the division office

When a link goes down for days, information ends up spread among people who cannot reach each other. At Wellington, the people on the mountain could see the snow on the slopes above them, the condition of the plows and the state of the trains. Officials at the division office could see the wider picture of men, coal, equipment and conditions elsewhere on the line, but they had no reliable way to learn what was happening at the siding. The passengers knew what they could see from the cars and what railroad employees told them. Without the wire, those pictures could only be compared when someone physically carried a message from one place to another.

The division superintendent, James H. O’Neill, was on the mountain directing the snow fight during the blockade. Krist’s book and other accounts describe him working at and around Wellington through much of the week. Details of his movements and decisions are where retellings vary, and I would send anyone teaching the case to Krist’s book and its cited sources rather than summarize them from memory. The general point holds whatever the details. An on-scene leader who has lost the link to headquarters has to choose between staying where the problem is and going to where communication is possible, and he cannot fully do both. Any decision he makes on the mountain is made without whatever headquarters knows, and any decision headquarters makes is based on information that is days old.

Popular accounts also describe passengers pressing railroad officials to move the trains, and describe disagreement over whether the siding, the tunnel or some other location was the safer place to wait. I have not examined those exchanges in a primary record, so I report them as what the accounts say. I am not going to judge, more than a century later, whether a different decision about where the trains stood would have changed the outcome. That is a hindsight question, and the decision-makers did not have the information that hindsight supplies. What I can say is that the people making those decisions were doing so without a working link to anyone who might have added resources, a second opinion or a different set of facts.

Messages on foot: the people who walked down to Scenic

With the wire unreliable, the most dependable way to move information off the mountain was a person walking. Accounts of the blockade describe several people leaving Wellington on foot in the days before the slide, passengers and railroad men among them, and making their way down the mountain toward Scenic. The descent involved deep snow and, by the descriptions in the popular histories, at least one very steep section that some of them slid or climbed down. Whatever each person’s reasons for going, every one of them carried information about conditions at Wellington to the valley.

A foot messenger is a poor communications channel even when the walk is safe. The message travels at walking speed through bad terrain, so it is out of date by the time it arrives. It goes in one direction per trip, so a question sent down the mountain may never get an answer sent back up. The content is whatever the messenger remembers and chooses to report, without the discipline of a written order or a logged message. And each trip puts a person into the same hazard everyone else was trying to avoid, which means the cost of a message was risk to a human being. When those are the only terms on which information moves, it moves rarely.

Families of the passengers faced the same gap from the other side. My understanding from the popular histories is that Seattle newspapers were reporting on the stalled trains before the slide, working from what reached the valley by wire when it worked and by people when it did not. I have not checked those papers myself, so I give that as what secondary accounts describe. A relative in Spokane or Seattle that week could learn that the trains were stuck, but probably could not learn much about a particular person aboard, because there was no path that let a passenger send a message out on demand. Modern reunification planning tries to solve that problem, and in 1910 nothing existed to solve it once the wire went down.

March 1: the slide and how word reached the valley

In the early hours of March 1, 1910, an avalanche came down the slope above Wellington and swept both trains off the siding and into the ravine of the Tye River below. Accounts describe the weather turning in the days before, with heavy snow giving way to rain and warmer temperatures, and many describe a thunderstorm with lightning during the night. Accounts also commonly mention that fires had burned the slope above the town in earlier years. I report those conditions because they appear consistently in the histories. I am not going to rank them or assign a cause, because weighing snowpack, weather and terrain for a slide in 1910 is beyond what I can do responsibly in an article.

The first rescuers were the people already at Wellington, railroad employees and others in the town, who went into the debris with shovels and their hands. Some people were pulled out alive. With the wire unreliable, the news of the slide had to leave the mountain the way information had left all week. Accounts describe survivors and railroad men making their way down to Scenic, where word could be passed on toward Everett and Seattle and help could start up the mountain. I am not going to give a time for when the first report reached the division office or the newspapers, because I have not seen that fixed in a record I trust, and the hour is easy to get wrong.

The communications consequence of that night is easy to state, even if the timeline is not. For the first hours, the only people who knew the trains were gone were the people on the mountain, and the only rescue force was whoever happened to be there. Outside help could start moving only after a person had carried the news down through the same snow and slide terrain, and then that help had to climb back up the same way. The isolation that had shaped every decision during the blockade now set the pace of the rescue.

The recovery: weeks of searching with no reliable link

Recovery went on long after any hope of finding survivors had passed. Rescuers and recovery crews came up from Scenic and worked in the ravine through snow, broken railcars, trees and slide debris. Accounts describe bodies being brought down the mountain by sled and then carried on toward Everett and Seattle for identification. By most accounts, the last bodies were not found until the summer of 1910, after enough of the snow had melted to expose them. A recovery that lasts months in terrain like that is a logistics problem, and logistics depends on communication: who is working where, what equipment is needed, what has been found, and who is still missing.

The same corridor that had failed during the blockade served the recovery, and line and track repair were part of the work. Until the wire was reliable again, coordination between the recovery site, the railroad’s offices and the mortuaries depended on the same slow relays that had carried news down after the slide. I do not have a detailed record of when reliable wire service to Wellington was restored, and I am not going to estimate one.

Casualty information moved through those relays as well. Families learned names from railroad announcements and newspaper lists, and those lists were assembled from reports that came in over days and weeks as bodies were recovered and identified. In my judgment, that process alone accounts for much of the variation in contemporary totals that later writers inherited, although I cannot point to a historian who has traced each variant number back to a specific list. For a modern agency, the lesson is that the casualty count during a long recovery is a communications product. It is only as accurate as the reporting path behind it, and it needs a single owner who publishes it with a time stamp and a counting basis.

What changed afterward, and what the record leaves open

The Great Northern renamed Wellington as Tye later in 1910, which is commonly described as an effort to separate the place from the disaster. The railroad built snowsheds over vulnerable sections of the line. In January 1929 it opened a new and much longer Cascade Tunnel at a lower elevation, which bypassed the Wellington site entirely. The old grade is now part of the Iron Goat Trail in the Mount Baker-Snoqualmie National Forest, where interpretive signs mark the site. Anyone who wants to understand how steep the slope is, and how exposed the siding was, will learn more from walking that trail than from any written description.

The disaster also produced legal proceedings, including a coroner’s inquest and lawsuits against the railroad. I am not going to characterize their findings or outcomes from memory, because legal conclusions are exactly the kind of detail that gets compressed and distorted in retelling. Anyone who needs them should go to the court and county records, or to a history that cites them directly, such as Krist’s book.

Several things remain open, or are at least not settled in any record I can cite with confidence. The exact toll is described above, with 96 as the accepted figure and higher numbers in circulation. The relative weight of weather, terrain and the burned slope in causing the slide is something I leave to avalanche specialists and historians. The hour-by-hour communications timeline, meaning when the wire failed, when it briefly worked again, and when the first report of the slide reached the division office, is something I have not seen established precisely enough to repeat. Those gaps do not weaken the communications lesson, because that lesson does not depend on the exact hour any message was sent. It depends on the fact that Wellington had one path to the outside, and that path ran through the hazard.

What to do at your agency

• The radio system administrator pulls the backhaul diagram for every tower site this month, marks on a map the physical route of each primary and backup link, and writes a list of every place where both share a bridge, a highway corridor, a single access road or a pole line.
• The emergency management coordinator adds one paragraph to the winter storm and flood annexes of the existing emergency operations plan. The paragraph names who is responsible for maintaining contact with stranded groups, such as motorists on a closed highway or residents on a cut-off road, and sets a scheduled check-in interval.
• The communications center supervisor tests the agency’s satellite phone or satellite messenger from the most isolated location in the county this month, and records whether it connected, how long the call setup took, and whether the battery was charged when the device was taken out of storage.
• The emergency management director asks the railroad that operates in the county for its after-hours emergency contact and its procedure for reporting a stranded train, and adds both to the railroad section of the existing contact list.
• The training officer adds an inject to the next scheduled winter weather tabletop in which the primary and backup links to one remote area fail together because they share a route. The exercise notes record how the group sets up contact with that area and who carries the first message.
• The public information officer writes one paragraph into the existing joint information plan. It states that casualty figures during a long incident will be released by a single named office, with a time stamp and the counting basis stated each time.

Takeaways

• Two Great Northern trains, the Spokane Local and a fast mail train, were held at Wellington near the west portal of the Cascade Tunnel for about a week in late February 1910 before an avalanche swept them into the Tye River ravine on March 1.
• The death toll most widely accepted is 96, the figure HistoryLink uses, and higher totals that appear in some accounts should be labeled as variants rather than treated as a settled competing count.
• The telegraph wire ran on poles along the railroad right-of-way, so the slides that closed the track also cut the only communications path, which is a classic common-mode failure.
• With the wire down, information moved mainly by people walking down the mountain toward Scenic, a channel that was slow and one-way and that put the messenger at risk on every trip.
• Decisions during the blockade were made without a reliable link between the people on the mountain and the division office, and the details of those decisions should be taken from sourced histories rather than retellings.
• Weather, terrain and an earlier burn on the slope all appear in accounts of the slide, and none of them should be presented as the established cause without a source that has done that analysis.
• For a working agency today, the durable lesson is to trace the physical route of every primary and backup link, and to plan contact with isolated groups before the storm rather than during it.