I've alwaysed liked the Atlas just due to the danger factor...( "You know John we haven't had much luck with the the Atlas rocket, this is going to be exteamly dangerous"....."I said we're ready!")
Thinking about the fins for the model flight, larger fins that stick out farther for stability, the real Atlas had no stablizer fins, I suppose the controll was all done by thrust vectoring?
Atlas had a pair of vernier engines mounted on the sides just above the separation plane between the booster section (housing the two outer engines and their pumps) and the main rocket vehicle body (the vehicle core). These verniers did several things... they helped with stability since they gimbaled (I'm not sure they'd have had enough "oomph" to provide ALL the stability so the booster engines and main engine probably gimbaled too, but I'm not 100% sure about that), they also provided roll control for the core once the outer two booster engines were jettisoned at "half-staging" which left only the center core engine to propel the vehicle... which also left it without roll control with no second engine to "gimbal against" for roll control (pitch and yaw could still be controlled by center engine gimbaling). The last thing the verniers did was to trim the velocity PRECISELY when the Atlas was flown as a nuclear missile... getting the velocity EXACTLY right is INCREDIBLY important in ballistic missiles, since the final velocity determines the range, or basically, the impact point... If the missile gets too much velocity, the bomb will "go long" and overshoot the target... if the missile shuts down a hair too soon and lacks a little velocity, the bomb will "fall short" of the target... of course the guidance system has to keep the missile on the exactly correct trajectory to prevent dispersions to the left or right of the target, and of course a trajectory that's too steep or too flat will also cause the bomb to fall short or overshoot the target, respectively. There was a pretty decent amount of uncertainty in the actual speed at shutoff, so on the nuclear missile version of Atlas, the verniers would continue to burn for as long as was required to increase the final velocity until it was trimmed perfectly for the exact range to the target. Assuming the radio guidance system kept the missile perfectly on trajectory, if the shutdown velocity was spot on, the bomb would hit the target pretty accurately.
Later missiles with inertial guidance and post-boost vehicles (missile buses on MIRV equipped missiles) improved the accuracy remarkably...
Really interesting stuff...
As for Atlas having "the danger factor", I think that's somewhat overstated... oh sure, Atlas (like everything else) had more than it's share of failures early on, but then again, they WERE the VERY FIRST ICBM's... rockets on that scale, power, and complexity had NEVER been attempted before. Karel Bossart's balloon tank construction, while seemingly flimsy due to the need to maintain the rocket in a pressurized condition at all times (unless it was resting securely horizontally in its special "half cradle" that provided external support to the missile's skin) might seem terribly risky, but really it wasn't... the main risk was with the propulsion, guidance, and sheer complexity of the rocket-- getting EVERYTHING together, 100% right, EVERY TIME it launched... the devil really IS in the details (as SpaceX found out today) and it wasn't just Atlas that suffered a lot of explosions on the pad or right after liftoff... Thor, Vanguard, Titan, etc. all suffered a, by today's standards, HUGE failure rate, mostly due to the fact that, at the time, virtually EVERYTHING was brand new and being done for the first time... it's hard to know exactly what will work and what won't work when you're breaking new ground and doing stuff nobody's ever done before. While it's true that one Atlas failed in flight when the Mercury capsule's adapter section failed and the capsule punctured the upper conical tank section, that problem was basically solved before it happened... that was the last remaining Atlas that hadn't been built with a "belly band" around the upper section to reinforce it, along with a stronger spacecraft adapter that was already in the works for all future missions.
The "balloon tank" and "stage and a half" method were REALLY FAR ahead of their time... I'd put those developments on par with Korolev's parallel booster staging concepts that led to the first practical liquid parallel staged vehicle, the R-7, Glushko's multi-chambered single-pump engines for higher thrust while overcoming combustion instability problems, and Von Braun's work in clustering, leading directly to Saturn I. It's truly a shame that we didn't take Atlas further, and design more vehicles based on the balloon tank stage and a half method... (remember that the "balloon tank" is actually pretty extensively used on practically EVERY rocket, to a degree... the propellant tanks are pressurized, to help push fuel through the lines to the turbopumps without cavitation, this is true, but ALSO because pressurizing the tank walls stiffens them up remarkably, allowing them to withstand FAR greater G forces and flight loads than they could EVER sustain on their own... the only difference with the balloon tank construction was that it had to remain pressurized to maintain its own shape and hold itself and the payload up while on the launch pad... which gives you a REMARKABLY lightweight structure! Remember too MANY early rockets that were 'self-supporting' on the pad failed in flight due to tank pressurization failures-- the gee forces and flight loads simply crushed them when tank pressure was lost or fell too low... ) Balloon tank construction gives some REMARKABLE performance advantages, as Centaur, the only other balloon tank vehicle, well proved... (along with liquid hydrogen propulsion).
The other thing is, the stage and a half concept, which is brilliant... especially when combined with the balloon tank construction to minimize the core vehicle weight... That is another technology that is underutilized... and which SHOULD be brought back, IMHO... The last vehicle to propose it was the National Launch System, which would have designed a new rocket using the Shuttle External Tank as a starting point for the tankage, mated to a new thrust structure and upper stages as needed, if needed. There was a design for a stage-and-a-half NLS booster that would eliminate the SRB's, and simply use SIX SSME's in a two-part thrust structure... the outer four SSME's would jettison along with the outer ring of the thrust structure partway through flight, after boosting the vehicle for a couple minutes to sufficient altitude and speed. The inner thrust structure, which supported the remaining two SSME's, would continue on under the power of these two engines either to orbit, or a suborbital injection point for core disposal in the ocean (requiring only a short circularization burn by the payload to inject itself into orbit by raising the perigee). The outer four SSME's in their half-stage thrust structure ring were eventually to be recoverable, parachuting inverted into the sea with flotation aides... (airbags) where they could be towed back to port by a ship.
Of course this derives somewhat directly from the work done on the S-ID proposal, which was proposed back before the shuttle was approved... it would have been a redesign of the Saturn V first stage, the S-IC, in which the thrust structure would be completely redesigned. It would have turned the S-IC into a stage-and-a-half to orbit booster... the outer four F-1 engines would have jettisoned after a couple minutes of boosting the core, leaving the center F-1 still firing to propel the core stage to orbit (or disposal orbit, depending on the payload). All this while STILL only burning kerosene and LOX, which has much lower ultimate ISP potential than liquid hydrogen propellant! The outer F-1's could have been made recoverable in the same manner as that proposed for the 1.5 stage NLS booster... parachute into the sea, nozzles up, and kept afloat via airbags. The payload of the S-ID booster would have been comparable to a space shuttle, for a FRACTION of the cost per flight... and it would have kept the possibility open for an HLV, since the S-IC was the first stage of the biggest HLV ever built... all that was needed were upper stage(s). S-ID could even have replaced the S-IC on the Saturn V, and significantly enhanced the flight performance of the Saturn V in that configuration over the S-IC boosted version...
Alas, the path not taken... instead of S-ID, we got shuttle, and we ALL know how that one turned out...
Later! OL JR