Part one of this series set out the communications findings from the 1993 bombing that were already written down and accepted before 2001. This part takes up the narrower and harder question of what the published investigations actually establish about radio performance inside the World Trade Center towers on 11 September 2001, where those investigations disagree with one another or decline to resolve a point, and what part of the record a working communications officer can use. I have kept to the primary documents where I could read them, and I say so explicitly at every point where I could not.

What a few watts is working against inside a steel framed tower

Start with the physical problem, because a reader who does not have it in front of them will read everything that follows as somebody’s error rather than as a property of the building. A portable radio held at chest height puts out a handful of watts into a short, inefficient antenna that is partly shadowed by the operator’s own body, and it has to get that signal out of a shielded volume built from composite floor slabs on steel decking, structural columns, core walls, elevator shafts, mechanical rooms, ductwork and the metallized coatings that go on modern glazing. Every layer in that stack takes energy out of the path, and the loss does not taper gently with distance the way it does across open ground, because it steps sharply according to exactly where the operator is standing and what is between the radio and the outside world at that moment. Two members three floors apart in the same stair can have completely different results, and neither of them has any way of knowing that from the front panel of the radio.

The direction that fails first is almost always the one from the portable outward. A fixed site or a building system has antenna height, antenna gain, a receiver with a good front end and, usually, a quieter noise environment than the one inside a stair shaft full of alarm cabling and elevator machinery. A portable has none of those advantages, so it is entirely ordinary for a member to hear the command post perfectly and be unable to answer, which is the worst possible arrangement, because the receiving end has no indication that the transmission it just made was not answered for a reason other than nobody being there. A radio check that is heard in one direction proves one direction.

Two other variables move the result and both of them get flattened in retellings. Coverage degrades with height when the serving site is on a rooftop several blocks away, because the geometry of the path gets worse as the portable climbs above the antenna’s useful pattern, and it degrades with depth into the building, because the core of a high rise is the most heavily shielded part of the structure and is also where the stairs and the elevator lobbies are. A portable that works acceptably standing at a tenant window on the fortieth floor may be unusable in the stair enclosure on the same floor, which is where the work is done.

None of this is a criticism of any department or any manufacturer, and it is the reason in building systems exist at all. A bidirectional amplifier or a distributed antenna system with radiating cable through the risers exists to move the transmitter and receiver that the portable has to reach inside the shielded volume, so that every portable has a short path to one point instead of a long path through the structure. The 9/11 Commission’s own statement of why fire department radios performed poorly in 1993 names propagation through the steel and concrete floors as one problem and the number of companies sharing a single point to point channel as a second, independent one, and that pairing matters for everything below, because a coverage device does not add talk paths and a channel that is being stepped on sounds to the person holding the radio exactly like a channel with no coverage.

The portables, the channels, and the digital radios that were not there

The fire department was operating analog portables of an older generation that morning, and the 9/11 Commission describes them in flat equipment terms in its chapter on the response in New York. The Commission writes that because these point to point radios had weak signal strength, communications on them could be heard only by other fire department personnel in the immediate vicinity, and it notes separately that the department had a dispatch frequency for each of the five boroughs, which were not point to point channels and could be monitored from around the city. The National September 11 Memorial and Museum’s catalogue entry for a recovered handheld radio used that day by the Chief of Department identifies it as a Motorola Saber, and the same entry states that in 2001 the department used analog point to point radios with low signal strength that historically performed poorly inside steel and concrete high rise buildings. I am not going to print frequencies, transmit power figures or a channel count, because I was not able to verify those numbers in a primary document while preparing this, and the place to read the equipment level description is the volume on the emergency response operations that the National Institute of Standards and Technology published as NCSTAR 1-8 in 2005, which the agency’s own abstract describes as addressing the technologies used during the operations as well as the guidelines and practices that governed them.

The channel structure matters more than the model number, and the Commission is specific about it. Companies operating in the North Tower kept their radios on tactical channel 1, which would be monitored by the chiefs in the lobby, and battalion chiefs who climbed would operate on a separate command channel. Units responding to the South Tower were advised to use tactical channel 3, according to the Commission’s account as summarized in the Gotham Gazette’s 2004 write up of the emergency response findings. Above those tactical channels sat the borough dispatch frequency, and separate again from all of them was the repeater channel inside the complex, which the record consistently calls channel 7. The designation of the command channel appears inconsistently across secondary accounts, so I am not going to attach a number to it here.

The department had newer digital portables in 2001 and they were not in service on 11 September, which is a documented sequence rather than a rumor, although the details are reported inconsistently and the disagreement is worth spelling out. Motorola XTS3500R digital handheld radios were issued in mid March 2001 and withdrawn within days after a firefighter’s Mayday transmissions at a house fire in Queens were reported to have gone unheard at the scene. A New York City Department of Investigation review is quoted in secondary sources as giving the issue date as 14 March 2001 and the Queens fire as 19 March 2001, a firefighting trade forum citing news coverage gives the issue date as 13 March with the radios pulled a week later, and FactCheck.org’s 2007 review of the same episode says the radios were yanked out of service two weeks after issue. Those accounts disagree on the issue date and on the length of the interval, I could not reconcile them against the Department of Investigation report itself, and anyone citing this sequence should read that report rather than any of the summaries, including mine.

What this establishes for the analysis is narrow and should be kept narrow. The department was working with analog portables whose in building limitations had been documented in its own after action record from 1993, so nothing about the radio performance inside the towers should be read as a surprise to the people who wrote either document. It does not establish that a digital portable would have solved the problem, because digital voice fails differently rather than less, holding quality until the error rate crosses a threshold and then dropping the transmission entirely instead of degrading into noise the way an analog signal does. The physics of the structure is indifferent to modulation.

The repeater: what the Commission recorded and what it did not resolve

Part one covered the installation, so the summary here is short. The Port Authority installed a repeater system for fire department use in the towers after the 1993 bombing, and the 9/11 Commission records that the system was installed at the Port Authority police desk in 5 WTC, to be activated by members of the Port Authority police when fire department units responding to the complex asked for it, and that in the spring of 2000 the department asked for an activation console for the system so that activation did not depend on that request being passed. What existed on the morning of 11 September, therefore, was a coverage device on a channel that had to be switched on by somebody at the start of an incident, with a console in the North Tower carrying more than one button.

The Commission’s account of what happened to it is unusually careful, and I am going to keep its hedges intact rather than smooth them out. One button on the repeater system activation console in the North Tower was pressed at 8:54, though the report states it is unclear by whom, and as a result of that activation communication became possible between fire department portable radios on the repeater channel. At 9:05, after the South Tower was hit, chiefs tested the system. Because a second button had not been activated, the chief on the master handset could not transmit, and the Commission states that he was also apparently unable to hear another chief who was attempting to reach him from a portable radio, either because of a technical problem or because the volume was turned down on the console, which was the normal setting when the system was not in use. Because the repeater channel seemed inoperable, the chiefs in the North Tower lobby decided not to use it. The Commission then writes that the repeater system was working at least partially on portable radios, and that firefighters subsequently used repeater channel 7 in the South Tower.

Notice what that passage does and does not settle. It settles that the channel carried portable to portable traffic through the repeater, and the Gotham Gazette’s summary of the same findings records that the chiefs who commenced operations in the South Tower lobby kept their radios on the repeater channel, unlike the commanders in the North Tower lobby, and that no chiefs outside the South Tower realized the repeater channel was functioning and being used there. Firehouse Magazine reported in 2004 that Commission investigators reviewed a 78 minute recording of fire department radio traffic captured through the Port Authority repeater, that the recording showed the repeater and the radios working, and that the chief in the South Tower lobby was able to speak with firefighters as high as the 78th floor. That recording is the strongest single piece of evidence on the question, and it is evidence about the South Tower operation specifically. The same Gotham Gazette summary records that from approximately 9:21 the ascending chief was unable to reach the South Tower lobby command post and that the lobby chief ceased transmitting on channel 7 at that time, so even on the channel that was working, the link did not hold for the whole operation.

What is not settled, and what I am not going to resolve, is the rest of it. The Commission gives two possible explanations for the master handset problem and chooses neither. It does not identify who pressed the button at 8:54. The Port Authority’s position, as reported by Firehouse Magazine in 2004, was that the repeater was in working order, which is an agency’s position rather than an investigative finding, and the Commission’s own wording is that the system was working at least partially, which is not the same claim. On the two other documents in the primary set, I want to be exact about the limits of what I checked: NCSTAR 1-8 contains the technical description of the repeater arrangement and is the document to read for it, and I am not going to paraphrase its conclusions from second hand summaries; the McKinsey and Company review of the fire department response, released in August 2002, is organized with a first part summarizing the key events including command and control and communications, and a second part of recommendations grouped under operations, planning and management, communications and technology, and family member support, and I was not able to verify a specific statement in it about the repeater’s status that morning. Whether the North Tower operation would have had a usable link if the master handset had worked is not answerable from these sources, and the honest answer is that it cannot be established.

The order to leave the North Tower and the several reasons it did not move everyone

This section needs more care than any other in the series, because a compressed version of it circulates widely and that version is not what the investigations concluded. The sequence, so far as I can support it, runs as follows. An order for units to leave the North Tower was transmitted before either tower came down, and the time given for it differs between accounts, with Firehouse Magazine reporting in 2004 that the senior chief who had taken over the North Tower lobby broadcast an order to leave the building at 9:28 on the two channels then in use, while an International Association of Fire Fighters video quoted in FactCheck.org’s 2007 article puts the same order at 9:32. Those two figures disagree, I could not reconcile them against the Commission’s own chronology, and the difference of four minutes does not change the analysis. The South Tower collapsed just before 10:00 and the North Tower about half an hour later, and the investigations state both times to the second, so take that precision from NCSTAR or from the Commission’s chronology rather than from this article. Evacuation orders were transmitted after the South Tower collapse, and the Commission’s account records that within minutes some firefighters began to hear evacuation orders over tactical 1, the channel being used in the North Tower.

Where people were at that moment is the first part of the explanation, and the Commission states it plainly. Just prior to 10:00, in the North Tower, one engine company had climbed to the 54th floor, at least two other companies had reached the sky lobby on the 44th floor, and numerous units were located between the 5th and 37th floors. The Commission also states that as units climbed higher their ability to communicate with chiefs on tactical 1 became more limited and sporadic, both because of the limited effectiveness of the department’s radios in high rises and because so many units on tactical 1 were trying to communicate at once, and that when attempting to reach a particular unit, chiefs in the lobby often heard nothing in response. A chief in the lobby is quoted in the same chapter saying “One of the most critical things in a major operation like this is to have information”, followed by the observation that not much information was coming in. Read as an engineering statement, that is the coverage problem and the capacity problem arriving together in the same channel, which is exactly the pair the Commission had already identified from 1993.

Reception was not uniform in either direction, and that is the part the compressed version loses. FactCheck.org, working from the Commission’s report in 2007, notes that two battalion chiefs on the 23rd and 35th floors not only heard an evacuation instruction on the command channel but repeated it to everyone they came across, so there were members well up the building who received the message and relayed it by voice. The same FactCheck analysis observes that factors beyond the radios were in play, including the absence of a constantly repeated Mayday order of the kind specified for a building believed to be in danger of collapse, which is that publication’s reading of fire department procedure rather than a finding of the Commission, and I present it as such.

The investigations give more than one reason and they give them together. Some members were assisting civilians who could not descend on their own, and the Commission’s record of the response includes accounts of exactly that work in the stairs. Some were at heights and in locations where a transmission on a point to point channel from the lobby was not going to arrive regardless of how many times it was sent. The information about what had happened to the South Tower was not uniformly available inside the North Tower, and the Commission’s staff work records that a fire department marine unit transmitted the South Tower collapse immediately and that, so far as the Commission could determine, that information did not reach the chiefs at the scene. The proposition that a tower of that construction would suffer a progressive collapse was not part of the operating assumptions on scene, and the basis for that finding is set out in the Commission’s chapter 9, which is where anyone relying on it should read it. What a working officer should take from this is that the explanation is multi causal in the primary sources, that radio limitation is one contributing cause among several rather than the whole account, and that any retelling which reduces it to a radio failure has dropped findings the investigators put on the record deliberately.

The call takers and the information nobody had a way to send them

The part of this record that matters most to a dispatch floor is the part that is usually left out, and it is the piece I would put in front of a new telecommunicator first. Civilians on floors above and below the impact areas in both towers called 911, and the Commission records that those calling from floors below the impact were generally advised to remain in place. That advice was consistent with standing high rise fire procedure as it existed at the time and had a defensible basis behind it, since uncontrolled self evacuation of a very large occupancy fills the stairs with people moving against the crews who are trying to climb, exposes occupants to the floors where the fire is, and historically has produced casualties of its own. I am not going to reproduce or characterize what individual callers said, and I would ask anyone writing about this to observe the same limit, because the analytical point does not require it.

The structural problem is the one to describe. The operators answering those calls did not have the picture that the chiefs in the lobbies had, and there was no mechanism by which they would have. In the arrangement the Commission describes, 911 was answered by police department call takers, with calls needing a fire or medical response transferred onward to fire department dispatchers, and the record includes callers being transferred more than once. None of those positions had a data or voice feed from either lobby command post, no position in the incident command structure had the job of pushing a status statement to the call taking floor, and the guidance the call takers were working from was written before the incident and could not update itself. The result is a system in which the people with the most direct contact with occupants were working from the oldest information available on the site.

The same disconnect shows up in the building’s own guidance channel and in the Port Authority’s answering points, which is the clearest evidence that this was structural rather than particular to one agency’s call takers. The Commission records that several South Tower occupants called the Port Authority police desk in 5 WTC, that some were advised to stand by for further instructions and others were strongly advised to leave, and that a Port Authority civilian group on the 64th floor of the North Tower received advice from the police desk at Newark Airport at 8:59 and from the desk in Jersey City at 9:10 that differed in its practical effect. On the building side, the Commission states that it is not known whether the fire department’s order to evacuate the South Tower was received by the deputy fire safety director making announcements there, and that at approximately 9:02, less than a minute before that tower was hit, an announcement over its public address system advised civilians generally that they could begin an orderly evacuation if conditions warranted, which like the earlier advice to remain in place did not correspond to any prewritten emergency instruction. Several answering points and one announcement system were each giving guidance drawn from different information, and none of them was connected to the command posts.

For a communications centre today the transferable question is short and can be answered this month. Somebody in your incident command structure has to own the job of telling the centre what the field picture is while the incident runs, at stated intervals, in a form the call takers can act on, and the centre has to know whose word changes its standing guidance and how that change is recorded in the log and in the recorded lines. If your answer to who does that is the incident commander when he gets a chance, you do not have a mechanism, you have an intention, and the interval between the field learning something and the call taking floor learning it will be whatever the workload allows.

Separate systems: what did not connect to what

Keep this section at the technical layer, because the command consequences are the subject of part three. Three police and fire organizations were operating radios in and around the complex that morning and they were operating on separate systems. The fire department was on its own tactical, command and borough dispatch channels as described above. The repeater channel inside the towers belonged to a system the Port Authority had installed, with its equipment at the Port Authority police desk in 5 WTC and an activation console in the North Tower, according to the Commission. The Port Authority police operated their own radio system with their own desks, including the desk in 5 WTC and desks at other Port Authority facilities that were in contact with occupants that morning. The police department operated its own radio system and also answered the city’s 911 calls.

At the radio layer the consequence is a negative statement, and the negative statement is the one I can support. I found no source describing a common tactical channel available in the portables of all three organizations that would have let a member of one department talk directly to a member of another inside the towers. Absent a shared channel, connecting two systems requires a gateway or a console patch at a fixed location, operated by somebody who knows both systems, with a decision already made about which talk paths get bridged and which stay separate, and nothing in the record I read describes such an arrangement in place for this incident. Where information did cross between the organizations, it crossed by a person carrying it, whether face to face at a command post or by telephone, at whatever speed that person was able to move it.

There is a capacity point that sits alongside the interoperability point and it deserves stating once. Even a shared channel is a single talk path, and the Commission’s finding on tactical 1 was that too many units on one channel were trying to communicate at once, which means that a bridge between two loaded systems can move a problem rather than solve it. The interoperability guide on this site makes the same argument from the governance direction, and the technical version is that a patch you have never loaded under realistic traffic is an untested assumption about capacity as well as about connectivity.

The figures, and what each source is counting

Several of the numbers attached to this day are quoted confidently and inconsistently, and a communications officer citing them in a class or a grant application needs to know what each source counts. The National Institute of Standards and Technology published its volume on the emergency response operations as NCSTAR 1-8 in 2005, with J. Randall Lawson and Robert Vettori named as its authors. The abstract of that volume states that the attack on the World Trade Center caused the deaths of 2,749 people, and that approximately 421 of them were emergency responders, a category the abstract draws widely enough to cover the fire and police departments, the Port Authority of New York and New Jersey and its police force, the security firms working at the World Trade Center, and volunteers who were in the area at the time. That responder figure is an aggregate across all of those categories and is explicitly approximate.

The three per agency figures most often quoted are 343 for the fire department, 37 for the Port Authority police and 23 for the police department, and they sum to 403 rather than to the NIST aggregate, because the NIST number counts a wider population including private emergency medical services, building security personnel and volunteers. I did not verify each of those three agency figures against that agency’s own primary published roll while preparing this article, so if you need one of them for a citation, take it from the agency’s published memorial roll or from the National September 11 Memorial and Museum registry rather than from a secondary source. One counting rule in particular explains why totals quoted years apart do not match: the fire department figure counts uniformed members who died that day, and deaths since then attributed to illness arising from the response are counted separately and have grown substantially, so a number published in 2002 and a number published this year are answering different questions.

The overall death toll has the same property. The figure in a 2005 federal report and the figure on a current memorial registry will not agree, because the toll was revised as the medical examiner’s office confirmed identifications and cause of death determinations over a period of years. The correct practice is to quote one source with its date attached and not to mix a component figure from one source into a total from another, which is the most common way these numbers go wrong in print.

Times need the same discipline. The collapse times of both towers are published to the second in the federal investigation and in the Commission’s chronology, and I have deliberately given them here as just before 10:00 and about half an hour later rather than reproducing seconds I did not verify in the source document. The 9:28 and 9:32 figures I gave earlier for the pre collapse order out of the North Tower are an example of the same problem at the small scale, since two published accounts give different times for what appears to be the same transmission, and the right way to handle that is to name both and leave the discrepancy visible rather than choosing the one that reads better.

What actually transfers to a working agency

The most common lesson drawn from this record is that somebody should have bought better radios, and that reading will not help your department, partly because the equipment question was already documented before 2001 and partly because it is not the part of the record that generalizes. Three things do generalize, and the first is that coverage inside a structure is a measurable property. You can walk a building with the portables your members actually carry, stand in the stair enclosures and the elevator lobbies and the mechanical spaces rather than at the windows, attempt a transmission to the position you would really use as a command post as well as to dispatch, and write down what happened with the floor, the location on the floor, the date and the radio identifiers. That produces a document you can act on, hand to a building owner, or attach to a preincident plan, and the reason most agencies do not have one for their tallest occupancies is that nobody was ever assigned to produce it.

The second is that a message which has been transmitted has not necessarily been received, and this is the single most useful sentence to take from the whole investigation. The Commission’s finding that chiefs attempting to reach a particular unit often heard nothing in response describes a condition every fireground officer will meet in a much smaller building, and the operational fix is not a better radio but a requirement for acknowledgement that is written into the way you run the incident. Emergency traffic gets acknowledged by the receiving unit by name, an evacuation order is relayed by the officers who receive it and confirmed back up the chain, a personnel accountability report is taken by assignment rather than by shouting into the channel, and a member who cannot reach command moves to a location where he can or hands the message to somebody who can. All of that costs radio discipline rather than money.

The third follows from the second and is the one that changes how a command structure operates. A command post with no means of confirming receipt is working on an assumption, and the assumption is invisible from where the commander is standing, because an unanswered call and a received order look identical at the console. If your plan for a high rise, a hospital, a jail or a large industrial site depends on an order reaching people who are out of reliable radio range, then the plan needs a relay position with a name and a location, or a runner, or an in building system whose service status somebody has confirmed at the start of the incident, and the confirmation step belongs in the plan rather than in somebody’s judgment on the day. The alternative is what the record shows, which is an operation where the lobby knows what it sent and does not know what arrived.

The dispatch floor side of this is the least expensive item on the list and the most commonly missing. Your call takers are giving guidance to people inside a structure your crews are working in, the guidance they give came from a policy written before the incident, and unless somebody has the specific duty of updating them while the incident runs, they will keep giving pre incident guidance for as long as the incident lasts. Part three of this series takes up the two command posts and what the investigations concluded about command and control across the response, which is where the consequences of the separate systems described above actually landed.

What to do at your agency

• Have your training officer or a company officer walk the two tallest buildings in your district that you would commit members into, with currently issued portables, attempting talk in from each stair enclosure on at least every third floor to both dispatch and the position you would use as a lobby command post, and record floor, location, result, date and radio identifiers on one page that goes into the preincident plan.
• Ask your fire marshal or building official, in writing, for the list of buildings in your jurisdiction required to have an in building emergency responder radio coverage system under whatever code edition your jurisdiction adopted, with the date of the most recent acceptance or annual test for each, and file the answer with the same preincident plans.
• For each in building system on that list, establish and write down who is responsible for it being in service, whether it is on continuously or requires activation, and how the first arriving chief confirms its status, naming the position that makes the confirmation rather than the agency.
• Have your radio system administrator demonstrate to you, on the air, exactly how an emergency or evacuation signal is transmitted on your system, then repeat the demonstration with a portable on the top floor of one of the buildings you walk tested and note whether the signal and the voice traffic that follows it are both received there.
• Add one item to the agenda of a meeting that already exists, your officers meeting or your communications committee, asking who in your incident command structure has the duty to pass a status statement to the communications centre during a working incident, at what interval, and where that duty is written down.
• Ask your communications centre supervisor for the current written guidance call takers give to occupants of a high rise during a fire, read it yourself, and confirm with the supervisor who is authorized to change that guidance mid incident and how the change is recorded on the logged lines.
• Write one paragraph into your existing high rise or target hazard procedure stating that an evacuation order is relayed and acknowledged by each operating unit by name, and naming the relay position or channel to be used when a unit above a stated floor cannot reach the command post.

Takeaways

• A portable radio transmitting a few watts from inside a steel framed high rise is working against the structure itself, the loss increases with height above the serving antenna and with depth into the core, and the path from the portable outward fails before the path inward, which is why a command post can be heard by members who cannot answer it.
• The 9/11 Commission describes the fire department’s radios that morning as analog point to point units with weak signal strength that could be heard only by personnel in the immediate vicinity, with tactical channel 1 in use in the North Tower, tactical channel 3 advised for South Tower units, a separate command channel for climbing chiefs, borough dispatch frequencies and the in building repeater channel known as channel 7.
• Newer digital portables issued in mid March 2001 were withdrawn within days after a Mayday at a Queens house fire was reported unheard, and published accounts disagree on both the issue date and the length of the interval, so the Department of Investigation report is the document to check rather than any summary.
• The Commission records that one button on the repeater activation console in the North Tower was pressed at 8:54 by a person the report does not identify, that a 9:05 test failed because a second button had not been activated and possibly also because a console volume control was down, that the chiefs in the North Tower lobby therefore decided not to use the channel, and that the system was working at least partially and was used on channel 7 in the South Tower.
• The record on the repeater is not fully resolved, since the Port Authority’s reported position was that the system was in working order while the Commission’s own wording is that it worked at least partially, and whether a usable link would have existed for the North Tower operation cannot be established from the published sources.
• The published investigations explain the failure of an evacuation order to move everyone out of the North Tower with several causes together, including members assisting civilians who could not descend, positions in the building where a point to point transmission was not going to arrive, and information about the South Tower’s condition that was not uniformly available inside the North Tower, and the single cause radio failure version that circulates is not what they concluded.
• Civilians who called 911 from floors below the impact were generally advised to remain in place consistent with standing high rise procedure, and the operators giving that advice had no feed from either lobby command post and no position in the incident structure whose job was to update them, which is the most directly transferable finding in this record for a communications centre.
• NIST’s NCSTAR 1-8 states 2,749 deaths including approximately 421 emergency responders across several categories, the three per agency figures commonly quoted sum to 403 and count a narrower population, and any figure you publish needs its source and its date attached because these counts were revised over years and count different things.