>Block lengths are something I've wondered about for some time. Without a certain formula spelled out by the feds, who or what sets the standard?

After skimming through 49 CFR Subpart B, 236, §204 specifies that "block length" must be based on the trains' stopping capabilities, not length in feet, yards, etc. This means it's up to each RR to define block lengths based on their operating considerations, and believe me they will also take liability in to account when doing this.

236 also specifies that the least restrictive signal indication which can be displayed by number-plated and controlled signals when the block is occupied is Restricting, not Stop as one would normally expect. On A-(controlled)signals, the SP used a flashing light (lunar or yellow) in a separate signal head attached to the side of the signal mast to enable a train to pass the A-signal without stopping when it displayed Stop (red). Other RRs used different means -- AT&SF used flashing red, as does the UP today.

>Modern signal designs seem to be moving toward 2 mile block lengths (or even longer) to accomodate heavier 143 TOB freight trains wherever possible.

Yes, 2 mile blocks (minimum). If you have access to a set of SP Common Standards, look up CS-2000, which shows SP standard signal layouts ("old style", the copy I have is dated 1932) and how protection is arranged in both directions between sidings. This has confirmed my decades-old suspicions about the SP's adoption of GRS's AP-B system lock, stock, and barrel. This makes sense because you don't want to have opposing trains leave two sidings and meet nose to nose in between.

However, since rules, plans, and standards can never be considered anything more than approximations of what to do in any given situation, no matter how good and comprehensive approximations they may be, CS-2000 gives nine standard layouts in order to accommodate the vagaries of local conditions (grades, curves, other operating conditions like busy yards, etc.).

The first six of these show opposing protection extending between sidings. The last three show it going only part way, and any one of these three would account for what the OP and his SP friends (and myself) have seen about those opposing signals staying yellow behind a train. The order of listing on CS-2000 is an implication of desirability. All we can say today is that probably one of these three schemes may have been used but not which one.

CS-2000 says this about block length: Distant signal - home signal = 3,000 feet; intermediate home signal = intermediate home signal = 3,500 feet; preliminary circuit: 2,000 feet. These values should be considered minimum distances. The "preliminary circuit" is what was known as a "lap" to us operating employees (short for "overlap").

Now things get a bit complex and confusing because we have to sort out Home Signal, Distant Signal, etc. A Home Signal is a signal at the entrance of a block, the indications of which govern the use of the block. A Distant Signal gives advance information about the position of a Home Signal.

If you go by one Home Signal which is Clear, you need to be able to stop for the next one if it's displaying Stop, because there is in all probability a train beyond it, either thousands of feet beyond it or merely a hundred feet or so. The worst case must always be assumed: the other train is 100 feet (or less) beyond the signal. Thus you need to have advance warning so you can stop before getting by that Home Signal. These principles apply to RR signalling in general to this day, virtually worldwide.

The SP, being a CS (Common Standard) road, like the UP, used lower quadrant style Bs semaphores. LQ semaphores can give only two indications: Clear and Stop. Advance information of a home signal's position is given by a separate blade, the Distant Signal, which also can show only two indications: Clear and Approach (yellow), "Approach" specifically meaning "you are approaching a Home Signal which is in Stop position". Home Signals carried red blades with a square end; Distant Signals carried a yellow arm with a fishtail end.

Sometimes a Home Signal had the Distant Signal for the next the Home Signal on the same mast, other times the were on two separate masts, separated by distances like those specified in CS-2000. For SP semaphore illustrations see: spcascades.railfan.net/SISKsig.html

The US&S H, H-2, H-2A, and H-5 searchlight signal mechanisms were drop-in replacements for semaphores. Unlike the LQ semaphores, they can display all three indications in one unit. So the Home Signals which also had Distant blade below the Home blade became simply one searchlight displaying all three indications: Clear, Approach, and Stop. The separate, stand-alone Home and Distant Signals were no longer necessary nor desirable, so they were either removed or converted into Home Signals governing new blocks, with control circuitry rearranged so there were also Distant Signals for the next Home Signal. What all this means is that after the style Bs were replaced with searchlights, there were no Distant Signals, except for very special applications/locations. Unless there was a D- on the signal's number plate, each searchlight signal became a Home Signal for the block immediately beyond it and a Distant Signal for the next [Home] signal.

Between 1990 and 2000 the old schemes have been pretty much replaced by the new ones -- those 2-mile long blocks.

This thread has dusted of a lot of cobwebs at my end -- some of this stuff first I read over 30 years ago. Railroads didn't install signal systems extensively until the WWI era and immediately afterwards, except perhaps for what is now called the NEC.

>This makes perfect sense to me given the longer stopping distances needed by unit coal or grain trains, which are almost always 140-143 TOB.

In the "SP days", the SP had Air Brake Rule 33, which stated that if you had a train which exceeded 80 tons per operative brake, you had to limit your speed to 45 mph, which is how the conditions of 236, §204 were met with those 1930s signal standards then in effect (even though the FRA didn't exist then, there was undoubtedly an ICC equivalent in effect). This was very important to keep in mind because, believe me, a 90 TOB train going 60-65 MPH simply doesn't want to stop, even on flat track with a full service reduction. I've tried blended braking with a freight, and it's quite effective. If handled correctly The Boss and his assistant in the Office never feel a thing, but with this particular freight (oil cans) I was unfamiliar with the concept.

>One thing I don't understand is why Santa Fe/BNSF prefer longer blocks in mountain territory where trains are already moving sufficiently slow that extra stopping distance isn't necessary. BNSF installed new CTC on the Needles Subdivision about ten years ago and extended the block lengths from about 1.9 - 2.5 to 2.5 - 3.1 miles just about everywhere except at sidings. The latter length was already common on the Cajon Subdivision.

Another thing I've noticed is that they seem to place all these new signals as close to grade crossings (public and private) as possible. Don't want the signal maintainers to have to hoof it.