For most of a century, the fastest and most dependable way to report a fire in an American city was to break a small pane of glass, pull a hook, and walk away without saying a word. The street fire alarm box sent a number, the number was the location, and the number arrived at the firehouse on a bell and on a strip of paper whether or not the telephone network, the electric utility or the person who pulled the box was still functioning. The engineering behind that is worth understanding, because the design principles did not become obsolete when the hardware did.

Bell towers, watchmen, and the problem Boston solved in 1852

Before the telegraph reached the fire service, a city reported fires by making noise. Watchmen stood in church steeples and municipal towers, and when they saw smoke they rang the bell in a pattern that identified a district, which brought companies out of their houses and into the street to look for the glow. The method worked in a town small enough that a man in a steeple could see most of it, and it degraded badly as buildings got taller, as smoke from industry became normal, and as the number of districts grew past what a bell pattern could usefully distinguish.

The fire alarm telegraph that went into service in Boston in 1852 is generally credited in the fire service literature as the first municipal system of its kind, and it was the work of William Francis Channing, a physician who had been thinking about the public safety uses of the electric telegraph, together with Moses Farmer, who built the apparatus. Their arrangement put signalling boxes on the street, connected them to a central office with wire, and had the central office retransmit the signal to alarm bells and to the engine houses. Channing published descriptions of the system, and the phrase fire alarm telegraph comes from that period.

The commercial history runs through John Gamewell, who acquired rights to the Channing and Farmer work for the southern states in 1855 and obtained the remaining rights a few years later. The Gamewell company, after absorbing or outlasting most of its competitors, became so dominant that in much of the country the street box and the brand name were used interchangeably. Published figures for how many cities had Gamewell equipment by the end of the nineteenth century vary between sources, and I would not treat any single count as settled, but the direction is not in dispute, because by the 1890s a municipal fire alarm telegraph was ordinary equipment in an American city of any size.

Inside the box: clockwork, a notched wheel, and four rounds

The critical design decision is that the box needs no electrical power of its own to send its signal. Pulling the hook releases a wound clockwork train, and that train turns a code wheel, which is a disc with teeth cut around its rim in a specific pattern. The teeth ride against a contact arm, and as the wheel turns, each tooth opens and then closes the electrical circuit that runs through the box. The box does not generate current or voltage; it interrupts current that the central office is already sending down the wire.

The pattern of interruptions is the box number. A box numbered 231 breaks the circuit twice, pauses, breaks it three times, pauses, and breaks it once, and then the wheel continues around and does the whole thing again. Standard practice was four complete rounds from a single pull, which gave the people at the receiving end multiple chances to read the same number correctly through noise, a weak contact or a distracted operator. Box numbering was not random, since the first digit or the number range commonly told the fire alarm office which circuit the box was on, so the number carried routing information along with location.

Some boxes carried more than the code wheel. Telegraph boxes included a key so that a fire officer at the scene could work Morse with the fire alarm office and ask for additional companies, and later generations put a telephone handset behind a second locked door, available to firefighters with a key and in some cities to the public. Keyless doors, breakglass panels and various anti-tamper arrangements came and went depending on how badly a given city was suffering from malicious pulls.

The important property of all of it is that a box is a fixed object bolted to a pole or a building at a surveyed location, so the code it sends is the address, which means the person reporting a fire does not have to describe where they are, spell a street name, know what town they are standing in or speak at all, and that last point became one of the arguments raised on behalf of deaf residents when cities later moved to take the boxes out.

The closed loop, and why a broken wire announced itself

A box circuit is a series loop. The fire alarm office pushes a small direct current out on a conductor, that current passes through every box on the circuit in turn, and it returns to the office, which means that under normal conditions current is flowing all the time and the office knows it. This is the opposite of the arrangement most people assume, where a device closes a switch to send a signal. Here the steady state is closed, and signalling happens by opening.

The consequence is supervision built into the physics rather than bolted on afterward. Any break in the conductor stops the current, whether it comes from a pole knocked down by a truck, a connection that has corroded open or a lineman opening a terminal without telling the fire alarm office, and the stop is immediately visible at the office on a galvanometer, a relay or a lamp. A crossed or grounded conductor shows up as a change in circuit resistance or as a reading on a ground detector. The operators took circuit readings on a schedule, and a circuit whose resistance had drifted got a technician sent to walk it before anything failed outright.

City practice reinforced that. Circuits were commonly run out and back so that a break in the middle could still be fed from both ends, which is the same reasoning behind a Class A initiating device circuit in a modern building fire alarm system and, for that matter, behind a ring topology in a radio system’s microwave backhaul. Dense cities moved the conductors into underground conduit, which cost a great deal and largely removed the ice storm and the delivery truck from the list of things that could take the system down. Lightning arresters sat at the office and at cable entrances, because a long run of aerial iron wire across a city is an excellent antenna for a nearby strike.

The tapper, the register, and the repeater in the fire alarm office

At the fire alarm office, the interruptions coming down the box circuit operated a relay, and the relay drove two things at once. One was an audible sounder, generally called the tapper, which reproduced the taps so that the operator on watch heard the number. The other was the register, a clockwork instrument that pulled a narrow paper tape past a punch or an inking stylus, so that each break in the circuit put a mark on the tape. The tape gave a permanent, time-referenced record of exactly what came in, which meant the operator did not have to trust a count made while three other things were happening.

That paper tape is the direct ancestor of every logging recorder in a modern communications center, and it was used the same way, since disputes about what time an alarm was received and what number was actually transmitted were settled by going back to the tape. Registers were also installed in engine houses, so that the company had its own record and its own means of reading a code it had missed on the gong.

The signal then had to get from the box circuit to the people who ride. In the earliest systems an operator listened, identified the number, and struck it back out by hand on the alarm circuits. The repeater automated that step, taking the incoming code and retransmitting it onto the bell and house circuits without human intervention, which removed both the delay and the transcription error. On the receiving end, a large gong in the apparatus bay struck the number, a tapper repeated it, and the register printed it. In the horse-drawn era it was common practice for the same alarm circuit to trip stall latches and turn on house lights automatically, so that the response began while the code was still being struck.

The company then looked up the number. Every engine house kept a running card or a box card file listing what each box number meant and which companies were due on a first, second or third alarm to that box, which is a pre-computed response assignment sitting on a card rack decades before anybody had a computer-aided dispatch system.

Two boxes at once: non-interference and succession

The weakness in a series loop is that every box on it interrupts the same current, so if two people pull two boxes on the same circuit within a few seconds of each other, the two code wheels turn independently and their interruptions interleave. What arrives at the office is a count that belongs to neither box and frequently resembles a third box somewhere else entirely. On a day when a large fire is spreading and several people reach for boxes at once, that is precisely the condition you least want.

The answer was the non-interference box, developed in the later decades of the nineteenth century and standard thereafter. Each box contains a mechanism that senses whether the circuit is already being interrupted by another box. If it is, the second box’s wheel is held, and it waits. The successive, or succession, feature then releases the held box so that it transmits its own four rounds cleanly after the first box finishes, which means both alarms arrive, correctly, one behind the other, without the operator having to sort anything out.

That is a genuinely elegant piece of engineering, and it is worth naming what it accomplishes in terms a radio person recognizes. It is collision detection and deferred retransmission on a shared medium, implemented in springs and levers, on a network where the endpoints have no power source and no clock in common. Trunked radio systems solve the same problem today with a control channel, a busy queue and a ready tone, and the failure mode when the queueing logic is wrong is the same one the fire alarm telegraph had before 1890, which is that two users transmitting simultaneously produce a message that is not either of theirs.

Why the system was extraordinarily reliable, and where it was not

The reliability came from a short list of properties that reinforced each other. The box needed no power, because the clockwork was wound by the act of pulling. The central office ran on its own batteries, charged locally, so a commercial power failure did not take the system with it. The signalling path was a dedicated municipal plant owned by the city, not a service leased from a company whose capacity was sized for normal traffic, which meant that a system in a city of a million people had exactly the same capacity on the night of a conflagration as it had on a quiet Tuesday. Supervision was continuous rather than periodic, so faults were found by the office rather than discovered by a citizen who pulled a box that did nothing.

The system also had no shared congestion mechanism to fail. In a telephone or cellular network, demand and capacity are pooled, so the event that makes everyone want to call is the event that prevents most of them from getting through. A box circuit carries one alarm at a time by design and hands the next one through immediately afterward, and the delay a busy circuit imposes is measured in the seconds it takes to finish four rounds.

None of that made it invulnerable, and the honest failure list is worth keeping. Aerial wire came down in ice and wind, and cities that had not gone underground spent significant money on line crews. Circuits could be crossed or grounded in ways that produced garbled or false codes, which is why the daily readings existed. The box told you a corner and not a building, so an alarm from a box on a block of tenements still required a search. Above all, the box conveyed a number and nothing else, so the fire alarm office could not tell an arriving company whether the call was a trash fire, a structure fire with people trapped, or nothing at all.

False alarms, the telephone, and the cost of copper in the street

Three pressures removed street boxes from most American cities, and false alarms were the loudest of them. An anonymous, free, silent device on a street corner that guarantees a fire department response is an obvious target, and by the 1960s and 1970s large cities were running apparatus hard on malicious pulls, with published rates that varied enormously between cities and between neighborhoods within a city. I am not going to quote a national percentage, because the reputable figures are city-specific and the sampling periods differ, but the fire service literature of the period is unambiguous that the false alarm burden was the argument that moved city councils. The injury and collision exposure created by responding was part of the argument as well.

The telephone was the second pressure. Once household telephone penetration was high, the citizen who saw a fire had a device in the kitchen that could report it and also describe it, and a caller who says the second floor is burning and two children are inside gives the fire department something a box code cannot. The arrival of 911 and then of enhanced 911 with automatic number and location information made the telephone path both easy to use and reasonably locatable, and the mobile phone finished the job by putting a reporting device on the same corner as the box.

The third pressure was money. A municipal fire alarm telegraph is miles of city-owned conductor, hundreds or thousands of mechanical devices exposed to weather and traffic, a staffed office, and a shop of technicians who understand direct current loop signalling and clockwork. That skill set was not being replaced as the technicians retired, and a public works director looking at a plant of aging underground cable can generate a very persuasive budget presentation. Cities removed boxes through the 1970s, 1980s and 1990s, usually in phases, and usually over objections from fire unions and from neighborhood groups.

New York is the well-known counterexample. The city moved to remove its street boxes in the mid-1990s, and a federal lawsuit brought on behalf of deaf and hard of hearing residents, reported in the Southern District of New York under the caption Civic Association of the Deaf v. Giuliani, resulted in an injunction against the removal on the grounds that the boxes provided access that the telephone did not. Check the reported decision itself rather than summaries of it, including this one, if the reasoning matters to your jurisdiction. New York still operates street boxes today, and the current inventory and its composition should be confirmed with FDNY rather than from older published counts.

Master boxes, radio boxes, and the hard-wired path argument

The part of the municipal system that survived most widely was the master box rather than the public box on the pole, meaning a box whose code wheel is tripped by a building’s own fire alarm control panel instead of by a person, putting that building’s address code onto the city circuit. Massachusetts is the best-known example of a state where municipal fire alarm circuits and required master boxes remain in ordinary use, and Boston continues to run a fire alarm office built around that model. Whether a master box is required, permitted or prohibited for a given occupancy in your jurisdiction is a question for your fire marshal and your local authority having jurisdiction, because the answer varies by state and by municipality and it has changed over time.

The national fire alarm code recognizes two ways a building system can drive a municipal box, generally described as local energy type, where the building system has its own power and operates the box mechanism, and shunt type, where the building’s initiating devices interrupt the municipal circuit directly. The shunt arrangement puts the city’s circuit into the building, which is why its use is narrowly limited. Look up the requirements in the edition of the code your state has actually adopted, since editions differ on this and adoption lags publication by years.

Where cities kept municipal signalling but could not keep the wire, the common replacement is the radio master box, in which the building panel trips a transmitter that reports over a licensed radio path to receivers at the fire alarm office, usually through a network of repeaters, with supervision by scheduled check-in so that a silent transmitter is treated as a fault. That preserves the useful properties of the old system, which are a dedicated path the fire department controls, capacity that does not collapse under public demand, and known supervision intervals, without the underground conduit. It substitutes a different set of exposures, including RF interference, receiver site power and the fact that the department now owns a radio system as well as a fire alarm system.

The argument for keeping some hard-wired or department-owned path is not nostalgia, and it comes down to shared congestion. Alarm communications today mostly ride commercial paths, and as carriers retire copper and traditional switched telephone service, the dual telephone line dialer that protected buildings for forty years is being replaced by cellular and internet communicators that depend on the same networks the public saturates during a major event. That is an acceptable risk for most buildings most of the time, and it is a different calculation for a hospital, a high-rise, a stadium or a chemical plant. If your city still has a municipal fire alarm circuit, the useful exercise is to ask what it still protects and what it would cost to keep, and to have that conversation before the last technician who knows the plant retires.

What to do at your agency

• Have your fire marshal or fire prevention officer produce a current list of every master box, radio box or direct-connect account your department monitors, with the building, the reporting path, and the date of the last test, and bring the entries with no test date to the next fire prevention staff meeting.
• Ask your alarm receiving vendor or your fire alarm office supervisor, in writing, what the supervision interval is for each communication path your department relies on, meaning how long a dead path stays undetected, and file the answer with your communications annex.
• Have your communications officer verify that the batteries, charger and transfer arrangement serving any remaining municipal fire alarm equipment are on the same preventive maintenance schedule as the dispatch center’s backup power, and add the load to the generator runtime worksheet if it is missing.
• If your department still operates municipal fire alarm circuits, have the shop supervisor write down the names of everyone currently qualified to troubleshoot them and their planned retirement dates, and put succession or vendor transition on the next budget planning agenda.
• Pull the running card or response assignment for one master box account and confirm it matches what your CAD actually dispatches today, because these records were frequently built once and never reconciled after a station move or an apparatus change.
• Add one paragraph to your existing communications continuity plan stating how the department will receive alarms from monitored properties if commercial cellular and internet service in your jurisdiction is unavailable or saturated, naming the fallback and who confirms it is working.

Takeaways

• The municipal fire alarm telegraph that went into service in Boston in 1852, developed by William Francis Channing and Moses Farmer, is generally credited as the first of its kind, and the Gamewell company later dominated the American market after acquiring rights to that work beginning in 1855.
• A street box needed no electrical power of its own, because pulling the hook released a wound clockwork train that turned a notched code wheel, and the wheel signalled by interrupting current the central office was already sending down the loop.
• Box circuits were series loops in a normally closed state, so a break, a ground or a crossed conductor changed the current or resistance and announced itself at the fire alarm office without anyone polling anything.
• The tapper let the operator hear the code, the register punched or inked it on a clockwork-driven paper tape as a permanent time-referenced record, and the repeater retransmitted the code automatically to bells and engine house gongs without a human transcribing it.
• Non-interference and succession mechanisms, standard from the late nineteenth century, held a second box until the first finished its rounds, which is collision handling and deferred retransmission implemented mechanically on a shared medium.
• The system resisted the failure mode that breaks commercial networks in disasters, because its capacity was dedicated and did not pool with public demand, though it was still vulnerable to storm damage on aerial wire and it could report only a box number and never the nature of the emergency.
• Malicious false alarms, household telephones followed by 911, and the cost of maintaining city-owned copper plant and the technicians who understood it removed street boxes from most American cities between the 1970s and the 1990s.
• New York’s removal program was halted in the mid-1990s by a federal suit brought on behalf of deaf and hard of hearing residents, reported as Civic Association of the Deaf v. Giuliani, and the city still operates street boxes, with the current count available from FDNY.
• The surviving argument for a department-controlled path is that alarm reporting over commercial cellular and internet service shares congestion and outages with the public, which matters more for a hospital or a high-rise than for an average building, and copper retirement is forcing that decision on jurisdictions now.