The astute student observing the tracks will often notice home signals which do not appear to be near, or protecting, switches, and often lack a yellow lens (for a diverging route) in their lower heads. The vast majority of these are holdout signals, enforcing the direction of traffic on a small or large length of track, as will be explained forthwith. There are some other cases (e.g., to protect drawbridges and to act in synchrony with a following home signal in certain cases of interlockings slightly removed from stations) but we will not consider them here.

(click to enlarge)

Consider the imaginary layout depicted in the single-line diagram above. If you do not already understand these symbols, please see Subway Signals: Single-Line Signal Diagrams. North and northward are to the right, South is to the left.

The diagram depicts a hypothetical interlocking and station, or (more likely) the two express tracks of a four-track station, which, although not at the physical end of its line, is with predictable regularity used as a the southern terminus of southbound trains. At such times, the trains come down from the North on Track 1, pull into either of the two "pockets" at the station, discharge and take on passengers, reverse direction, and pull out to the North on Track 2. At other times, trains ("through traffic") simply proceed straight through, southbound on Track 1 and northbound on Track 2.

Clearly, if the station is in use as a terminal, and trains are occupying either station pocket for extended periods of time (as usual at a terminal), through traffic can never pass them, so this mode of operation would most likely occur on the express tracks of a four-track line, in which case through traffic can proceed on the local tracks. There are many such places in the system ---- Atlantic Avenue on the IRT Brooklyn Line and 34th St. on the IND 6th Avenue Line are two that come to mind. (The interlockings at both of these places are singularly complex, in order to facilitate the smooth exchange of trains between local and express tracks in addition to the terminal operation, but since we are only concerned with the latter, this simplification will serve our needs).

The above layout also represents a simplification over the realities it models in some other ways, e.g., the lack of station time control to cut back control lines allowing trains to close in on each other, and the shortness of the control lines (in the absence of grade time control to force a speed so low as to justify such short control lines (including, typically, a reverse-direction grade time signal back-to-back with B6)).

The home signal 4, in conjunction with its approach signal, 2, and the marker signal, 8, are used to convert the normally northbound Track 2 pocket between signals 8 and 12 into a terminus for southbound trains. The interlocking will not allow signal 16 to be cleared over switch 15 reverse (or 20 over 13 normal, that is, to an exit at 12), unless all of the levers (or other control apparatus) for signals 2, 4, and 6 have been reset to force them to indicate stop.

When the northbound pocket is being used as a terminal, both signals 4 and 8 are impassable: their train stops are both in the tripping position, preventing trains from passing southbound past 8 or northbound past 4. Both signals are absolute: Neither can be keyed by, nor can the home signal 4 (at such times) display a call-on. In fact, signals 4 and 8 are tightly coupled: only the calling of signal 4 (i.e., setting its lever or other control apparatus to a state in which it can clear) can drive the train stop of 8 into the "clear" position; even its "back-to-back" approach signal A6 cannot clear 8's stop, as a marker signal is said to be of a higher priority (authority, urgency) than an approach signal. The joint action of these two signals and their stops enforces a complete isolation between the southbound pocket and the rest of the northbound track. 4 is called a holdout signal, as it holds northbound trains out of the southbound pocket. The absolute marker 8 acts as a "virtual bumper" for terminating southbound traffic.

The approach signal 2 serves the same function with respect to 4 as do other approach signals with respect to their home signals in protecting switches (e.g., as 6 works with 12 to protect switch 13B), i.e., to ensure that trains are not just stopped at the home signal, but that no trains may barrel past ("overrun") it, stop or no stop, and carry on by momentum: the approach signal ensures that they must already have been stopped at 2 to even approach 4 when the northbound pocket is in use as a terminal. See the explanation of approach signals for more detail.

Note that this arrangement is only necessary on the northbound track, because there are, or supposed to be, no northbound trains on the southbound track. While the system is not designed for "rogue trains" going the wrong way without permission on the wrong track, it nonetheless prohibits such trains from entering such track in the wrong direction in the first place. For example, observe the (quite typical) note on signals 10 and 12, "NO MOVES TO TK 1". The interlocking will simply not permit 12 to be cleared over 15 reverse, or 10 over 13 normal (i.e., not offer exits at 16), even by call-on. The missing-light lower heads of these two home signals reflect these deliberate limitations.

Note furthermore that this arrangement (i.e., signals 2, 4, and 8) is only necessary because of the need to terminate trains on two tracks, that is, to implement a two-pocket terminal. Were there only a need to turn around an occasional southbound train, say, in emergencies, the Track 1 pocket alone would suffice, with trains exiting northbound past signal 10 to Track 2. No special protection is needed for a southbound train to enter the southbound pocket. Nor is any (other than 18 and 16 protecting the switch 13A) needed for it to turn around and proceed North over switch 13 to Track 2. For that matter, this interlocking can turn around northbound trains at signal 20, at no additional signalling cost (other than that signal) at all. It is thus clear that a trailing-point crossover (e.g., 13) is far easier and cheaper to signal and easier and safer to operate than a facing-point one (e.g., 15) for simple turnarounds. In fact, the only place in the system where a facing-point crossover is so used is at Prospect Park on the Franklin Avenue Shuttle, most likely because the southbound approach to the southbound pocket is the infamous curve which was the site and cause of the bloody 1918 Malbone St. wreck.

In older interlocking plants, 20 would be a dwarf signal, and perhaps 10 as well.

Traffic Locking (Traffic Control)

Holdout signals are a technique for managing bidirectional traffic on a piece of track completely within one interlocking, under the supervision and control of one tower operator. When a lengthy piece of track runs between two interlocking towers, and must, however infrequently, but predictably, support controlled motions of trains in two directions, a technique known as traffic locking (or traffic control) is employed (by many railroads, not merely the New York Subway system).

Consider the single line diagram below, representing a long track, likely the southbound track of a pair, running between the interlockings at the fictional A St. and B St., each under the control of their own tower operator. As can be seen, the track is signalled only in the southbound direction. This track might be part of an underwater tunnel, normally used in the southbound direction, but on rare occasion, when the other tunnel tube is being repaired, run in the northbound direction as part of single-track operation, but not frequently enough to merit full signalling in that direction. In single-track operation, trains are let into a single track in alternate directions, with only one train in the tunnel at once, to rule out the possibility of collisions in the absence of signals. Such operations are quite common in rapid transit systems.

(click to enlarge)

Before we describe the semantics of traffic control, please note that the signals 2 and 4 at Tower A have control lines extending all the way to the signals controlling entrance to the track at Tower B. This ensures that no more than one train can possibly be in the long section unsignalled in the reverse direction, for that very reason (i.e., signal protection against rear-end collision is lacking). Typically, not even call-on moves are permitted in this circumstance.

The heart of traffic control is the stipulation that A and B towers must agree on the nominal ("official") direction of traffic on the traffic-controlled section before either may let trains into it. Furthermore, the direction may not be changed once trains have ingressed upon it (i.e., it must be vacant before direction can be changed). Each tower has a control called a traffic lever associated with the control of direction on this track. At an (old) electromechanical or electropneumatic machine, this is an actual lever on the interlocking machine. At all-relay / pushbutton interlockings, it is more typically a knob on the track plan that can be turned one way or the other, or a miniature lever. In any case, it is spoken of as though it were an actual lever. At Tower A, it is lever 7, and at Tower B, lever 13. The traffic levers at both interlockings must agree before trains can enter the controlled track.

The interlocking which feeds trains into the controlled section is known as the entrance end, and the other, which receives them, the exit end. Acting as the exit end requires no coordination with the other tower. For example, Tower A can clear signal 6 for a move to 2 or 4 at any time without any safety considerations involving the controlled track or Tower B. Even if Tower A is the entrance end, a reverse-direction (southbound) move entering at 6 is safe. The same is true for northbound moves entering at 12 at Tower B. Being the exit end is easy.

In order for the tower currently the exit end to become the entrance end, the tower at the other end, currently the entrance, must first throw its traffic lever to attempt to make itself the exit end. Then, if there are no trains in the section, the tower which is the current exit end can throw its traffic lever, and make itself the entrance end, and at that instant, the new nominal track direction and roles become effective.

Traffic control prohibits ("locks out") exits onto the controlled section at the interlocking which is not the entrance end. For example, let us assume that (both interlockings agree, and their traffic levers are set) that traffic is southbound on the controlled section: Thus, B is the entrance end, and A the exit end. In this condition, interlocking A (specificially its Traffic Lever 7) will refuse to allow signals 2 or 4 to be called (e.g., their levers to be drawn from the normal position to permit clearing). Furthermore, traffic control prevents the changing of direction unless all such routes have been cancelled and reset. Thus, in the same condition, let us suppose that Tower B had cleared its Signal 14 for a southbound route onto the controlled section; Tower B cannot begin the process of switching traffic direction by becoming the exit end until Signal 14 is set to stop and reset: Signals 14 and 16 will lock traffic lever 13 in the southbound position which was necessary to achieve their call in the first place.

If there are approach signals of either interlocking in the controlled section, they, too, participate in traffic locking, and must be reset before the direction changes against them, and, conversely, require traffic direction to agree with them before they can be called.

Traffic locking is sometimes used in conjunction with holdout home signals (see above), in which case the traffic-controlled section really starts at the holdouts.

True Bidirectional Track

Often, tracks are used heavily in both directions, and fully signalled in both directions. The most common examples of this are the "center express tracks" common on the IRT in the Bronx and other elevated lines in the outlying boroughs, which change direction for prevailing rush hour traffic. Consider the following:

(click to enlarge)

Here, there are copious signals in both directions. Thus, unlike the case of single-track operation, many trains (in the established, nominal direction of travel) can be in the controlled section at once, fully protected from each other. The control lines of signals 2 and 4 at Tower A do not extend to Tower B, but only the distance necessary to protect the rear end of the next train (i.e., the normal determinor of control length).

Traffic control is used in such cases exactly as described above. The presence of the additional automatic signals does not affect the conditions for assigning and changing track direction, nor the cooperation and coordination sequence required between the two towers. In this case, however, the traffic control levers will affect the automatic signals in the controlled section: in the absence of trains, all of the automatic signals in the "wrong" direction will be at stop (red), but with their train stops "driven" (to the clear position) to permit reverse-direction traffic, while the signals in the established direction of traffic will be clear and operate as normal automatic signals. In effect, the track becomes "unsignalled" in the direction opposite to the established direction, and no trains are permitted to enter in that direction. (On the unresignalled BMT, currently only the West End line as of this writing (6 June 2001), the signals in the "wrong direction" actually go dark.)

Bidirectional traffic-controlled signalling is also used on some tracks that are so often single-tracked (in repair contingencies) that such has become necessary. The two tracks of the Brighton Line between Prospect Park and DeKalb Avenue, straight through the Montague St. Tunnel Line to Lawrence St. Interlocking, are an example. In recent years, all two-track lines constructed or resignalled (e.g., the surviving piece of the Myrtle Avenue El, the Archer Avenue Extension) have been signalled in this style.

Page Credits

Descriptions and graphics on this page (unless otherwise noted) are Copyright © 1997-2002 Bernard S. Greenberg (contact).