The Gemini spacecraft, often given as an example of a manned space capsule with no launch escape system, or LES (it used individual ejection seats instead), did, in fact, have an LES, although not one useable for the entire ascent (hence the ejection seats); for launch aborts above 70 kilofeet altitude, the crew capsule’s four solid-fuelled “retrorockets” (in reality large launch-escape rockets; the Gemini capsule’s RCS had more than enough Δv to deorbit the capsule onto a healthy reentry trajectory on their own, and the launch-escape rockets were fired merely to use them up so that they wouldn’t explode during reentry in the vicinity of the capsule and potentially cause damage) would fire to get the capsule away from the possibly-exploding rocket, after which it would parachute back to earth and splash down in the ocean.
The Gemini B capsule, used as part of the cancelled Manned Orbiting Laboratory (MOL), used a similar launch escape system, except with six escape rockets instead of four, enabling it to be used for off-the-pad and low-altitude ascent aborts and to get the capsule further away from the launch vehicle than would have been possible with the four-rocket system; this was necessary due to the use of the much-more-explosive Titan IIIM launcher, which included two large SRBs not found in the Titan II GLV launcher used for the NASA Gemini missions. Any MOL launch abort would therefore have involved the crew capsule separating from the launch vehicle, rather than leaving the capsule attached to the launcher and merely ejecting the crew.
However, even then, early aborts (off the pad or in the first 31 seconds of ascent) would still have seen the crew ejecting from the capsule after its separation from the launcher:
...If an abort was required before liftoff or up to 31 seconds later, salvo-firing all six retrograde rockets simultaneously would rocket the Gemini-B to a safe distance from the exploding booster, allowing the pilots to eject and land under their personal parachutes. [...] In any abort from 31 seconds to separation of the solid rocket boosters, the pilots would not eject but would stay in their Gemini-B capsule through re-entry and splashdown. [John B. Charles, “A Jones for MOL #11: The Retroactivity of MOL (Part 1)”]
This makes less than no sense to me, as the whole point of escaping from an exploding rocket is so that the crew can land safely, and getting the capsule away from the exploding rocket would seem to accomplish that goal; the Gemini and Gemini B were fully capable of landing safely with crew inside (as can be seen by the fact that MOL launch aborts after T+31 would have involved the crew landing with the capsule), and the extremely-powerful Gemini ejection seats would have had a very high risk of causing severe or fatal injuries to their occupants had they ever been used.1
Can someone explain why off-the-pad and early-ascent MOL aborts required the crew to eject after the capsule separated from the launch vehicle, seemingly endangering the crew for no reason?
1: And bear in mind that the Gemini ejection seats were very-high-risk by the standards of 1960s ejection seats, a time when normal ejection seats had about a one-in-three risk of killing or crippling their users even if used properly.
