Red, Yellow, Green

The color aspects of subway signals are vaguely similar to those of street traffic lights -- red means "don't go, but stop," yellow means "slow down," and green means "go". The similarity, however, ends there. Green does not just mean "go", but certifies that the next signal, the one after the green one, doesn't say "stop". Yellow is even more different in meaning: While a yellow street traffic signal means "slow down, because this signal is in the process of changing to red" (which many motorists, of course, interpret as "speed up so as to pass it before it does"), a yellow subway signal means "slow down, (most often) because the next signal already is red, and you must slow down and proceed with caution before reaching it."

While street traffic signals usually go from green to yellow to red, subway signals usually go from red to yellow to green.

Approach and Automatic Signals

Except on the rapidly vanishing unresignalled portions of the IRT, approach signals and automatic signals cannot readily be distinguished from each other by the way they look nor the way they are treated: their aspects (combinations of colored lights they display) and their indications (what the train operator is instructed to do by them) are identical. They are single-headed (one case with lights in it, although occasional single lights appear below) signals, and the only kind of block signals that appear on revenue trackage other than the home signals at interlockings (loosely, places where tracks meet, join, and switch). They both have train stops ("trip arms"). Almost all the signals you will see riding the subways are automatic signals.

Automatic signals are controlled solely by train motion, i.e., the occupancy and vacancy of lengths of track. Sometimes, timers (controlled by train motion) are involved as well, to ensure that trains are operating slowly enough before the signal will clear (change from red) all the time (grade time (GT)) or when trains are ahead (station time (ST)). The distance down-track ahead of the signal, in which if there is any obstruction the signal will indicate "stop", is called its control length, and is the single most important concept in signal design. Control lengths always overlap beyond the next signal.

Approach signals are just like automatic signals, but they are also in part controlled by an interlocking tower, where the tower operator can (and must, in certain circumstances -- see Interlocking) force the signal to indicate "stop" (red) even if there are no trains ahead of it (which is why these signals exist in the first place). Modern approach signals have a white number plate on the side facing the train showing the number of the controlling lever at the relevant interlocking. They often come in pairs, triplets, or quadruplets controlled by the same lever, with deviously ingenious overlapped station time control, in a station's track berth approaching an interlocking. Although also interlocked, approach signals are different from home signals in that while the latter are absolute (may not be passed when red), the former are not absolute, and may be passed under "stop within vision" rules as described below. If you really want to know why approach signals are needed and exactly what function they serve, please see Why are there approach signals?

Except on the "old IRT" (as marked), automatic signals and approach signals at "stop" (red) can be passed, (i.e., their train stops can be driven down by the train operator) by automatic key-by (AK) or <a manual key-by, although the latter is obsolescent and present only where marked. On trackage still using the "old IRT" system, automatic signals are absolute unless explicitly labelled with a yellow AK (automatic key-by), or a sign with a red K (manual key-by).

**Approach and Automatic Signal Aspects and Indications**
(Distant aspect) Proceed-- the next ("distant") signal is clear (not red).	Proceed with caution, prepare to stop at next signal.	(Stop, with automatic key-by) Stop, operate automatic stop automatic release, then proceed with caution, prepared to stop within vision expecting to find track occupied, a broken rail, or other obstruction in the block	(Stop, with manual key-by) Stop, operate automatic stop manual release, then proceed with caution, prepared to stop within vision expecting to find track occupied, a broken rail, or other obstruction in the block
(Lunar white "one shot" GT aspect) Approach at posted speed, expecting this signal to clear if you do (be tripped if you fail)	("Two-shot" GT aspect) Approach at posted speed, expecting the next signal to clear, making this one green, if you succeed.	Approach at posted speed, expecting a diverging route at end of block ("Two shot" GT, same as previous, but the next signal is a home signal which will clear with time to a diverging route.)	(ST (Station Time) aspect) Approach at allowable speed indicated by illuminated number, expecting signal to clear if you do (tripped if you fail).

Marker Signal Aspects and Indication

Marker Signals are the subway equivalent of "do not enter." They unconditionally deny access, almost always in the reverse direction to usual traffic to interlockings that only have logic and signalling for traffic in one direction (e.g., De Kalb Avenue on the BMT in Brooklyn). They never change their red aspect. They often have train stops (automatic "trip arms"), and are absolute (may not be passed when at "stop", which, for a marker signal, is always.).

While extremely boring to watch, marker signals are nonetheless important in the science and practice of interlocking, where they represent necessary "entrances" to interlockings from which trains never enter. If this sort of thing interests you even slightly, please check out NXSYS, Signalling and Interlocking Simulator.

**Marker Signals**
Stop and stay	Stop and stay

Train Order Signals

Train order signals instruct the train operator to call an appropriate dispatcher or other authority about a change of plans. These signals have decreased in importance since the advent of full-time radio contact with all trains.

**Train Order Signal**
Telephone for orders, stop and stay -OR- no orders to be received, proceed

Repeater Signals

A repeater signal is merely a "carbon copy" of a block signal (of any type, home, automatic, approach, etc.) on the left-hand wall of a tunnel around a clockwise curve. It displays an identical aspect to the signal it "repeats", which is across the tunnel on the right-hand wall, and exists solely to be visible to the train operator as he or she brings the train around the curve, and cannot see the "real" signal. The number plate of a repeater signal always has the letters REP instead of a line designation.

**Repeater Signal**
Next signal on cab side displays same aspect (automatic and home repeater signals are shown here).

Miscellaneous Non-Block Light Signals

The "switch stand" signals below, used to show which way a switch (the technical term for a 'switchtrack', i.e., the junction of two tracks) is thrown, are used only with non-interlocked switches, that is, those controlled by direct physical manipulation and not an interlocking tower. Although common in some other transit systems, especially for little-used switches, non-interlocked switches are extremely rare in New York, and exist only in a small number of yards (i.e., not on revenue trackage at all). Because they are not coordinated with signals, they are innately unsafe, which is why interlocking, the logically constrained coordinated operation of switches and their surrounding signals, was invented in the first place. Gap fillers are mechanical fingers that extend out from a concave station platform on a curve to meet the middle door of a car (they are employed only on the (former) IRT division) so passengers don't fall into the gap. They have been used at 14th St. on the Lexington Avenue line and at South Ferry (outer loop).

**Switch stands, gap filler signals, holding lights, and bumpers**
Switch set for same track and/or derail in non-derailing position, proceed at allowable speeds	Siding or yard switch set for diverging route and/or derail in non-derailing position, proceed at allowable speed	Main track switch set for diverging position and/or derail in derailing position, stop and stay
Gap fillers out, stop and stay	Gap fillers in, proceed	Train starting or holding signal; at a terminal, illuminated to dispatch - anywhere else, illuminated to hold.
Main line bumper stop
Yard bumper stop

Why are there approach signals?

As you may have read, approach signals are visually indistinguishable from automatic signals, but, like home signals, are controlled by an interlocking tower. You may wonder why they exist at all. The following diagram and explanation should motivate the need for approach signals. In this diagram, the lines represent tracks, and the little circles with numbers represent signals. Trains go from the left to the right. These signals are roughly 200' apart. Only relevant signals are shown. The colors represent the aspects of each head of each signal. The switch is set for Track 2.

Even though there are no trains on the diagram, it should be clear why the home signal labelled "4" ("signal 4", or "4 signal", as signal design and operating personnel often say) is at "stop and stay", "red over red". As the switch is set for Track 2, trains may not enter from track 1, lest they be derailed, damage the switch, or both, or even possibly collide with a train coming in from track 2. If the switch is set for track 2, passing signal 4 is absolutely prohibited, in any way at all in any circumstance, so it must be (absolute) "red over red" (please see home signals for home signal aspects and indications.)

However, merely signal 4 being at "stop" is not sufficiently safe. Were the switch as shown, and a train to come in on track 1 at full speed and hit signal 4, it would indeed be tripped by the train stop of signal 4, and eventually come to a halt. But trains requires many hundreds of feet to stop, and by the time the train stopped, an accident would have already happened when the slowing train encountered the switch.

Therefore, when a switch is set the "wrong way" for a given track, enough signals in advance of the switch on that track must be red so that if a train at full speed were tripped by the train stop of the first of them, it would grind to a halt before getting near the switch. For signal 4 to succeed in its role of protecting the switch when it is set for the other track, it must be sure that trains are slow or stopped when approaching it. That is why signals A2 and B2 need to be controlled by the interlocking tower -- they must be set and guaranteed to be red before the switch can be reversed (set to track 2). They are always controlled together, by the same lever (lever 2, see interlocking), and hence are designated A2 and B2. They make sure that trains cannot approach signal 4 at speed when the switch is set reverse.

You might be asking yourself, "Why can't you move signal 4 far back enough such that this is not the case?" Well, the placement of signals is determined by how close the designers want trains to approach each other, and in this case, how close trains can get to the switch, so it has to be that close, and more signals are required. There is nothing wrong per se with a train coming all the way up to signal 4 when the switch is set against it, just not at speed; in fact it is desirable -- trains should be allowed to go as far as they safely can. That is why (except on that ol' "old IRT") A2 and B2 are not themselves home signals -- even when they are red, it is safe to fully stop and proceed past them slowly with automatic key-by. Having stopped at an approach signal, a train cannot build up enough speed to endanger the switch if tripped by the home signal. This constraint, too, is paramount when the placement of approach (and home) signals is designed. As you can see, the safe placement of signals requires balancing of many factors and very substantial expertise.

Note that it is not adequate to simply make the approach signals red when the switch is thrown -- like home signals, they must be set and ensured to be red before the switch can be moved. And if there is already a train within or approaching them at the time the tower operator decides to reverse the switch, even that can be seen to be insufficiently safe ("the proof (and the solution) is left to the reader"). Please see interlocking for a further discussion of these issues, and investigate NXSYS, Signalling and Interlocking Simulator if you find that this whets your appetite for even more laborious technical detail.

Subway Signals: Approach, Automatic, and Marker Signals

From nycsubway.org