Launch Complexes 39A, 39B, 39C, and 39D
When Launch Complex 39 was initially being designed, it was not yet decided exactly how the U.S. would be getting to the moon, how many launches would be required for a lunar mission, or how large the rockets would be. As such, the LC-39 designers erred on the side of caution, designing a launch complex which would be able to accommodate any reasonable launch rate and could be expanded for an even higher capacity.
Prior to July 1962, the lunar mission "mode" had not yet been decided. The leading candidate was earth-orbit rendezvous, where two or more rockets would be launched, rendezvous in earth orbit, assemble the lunar spacecraft, and proceed to the lunar surface; this method would require a fairly high launch rate per mission.
I have a pamphlet dating back to perhaps early 1963, which was a time of real excitement and optimism about the program. The pamphlet touts that "[f]our launches of the Saturn V are currently scheduled for 1966, and six are planned for 1967." After the first launch in "early 1966", operational launches would occur "about two years later"; goodness only knows what the operational launch rate would be.
With such a high launch rate, a large launch complex with several launch pads would certainly be required, and this pamphlet shows such an early concept, where LC-39 had four launch pads (adding LC-39C and LC-39D to LC-39A and LC-39B, the two pads which were actually built). Note the "kink" along the route to the north-most pad; presumably, this would allow the eventual addition of yet another launch pad! LC-39E, anyone?
And LC-39E is actually depicted in this diagram dating back to February 1963, with LC-39D and LC-39E being shown in dotted lines and labelled "Future":
In addition to the Saturn V (at the time known as the "Advanced Saturn" or the "Saturn C-5"), it also shows the proposed Titan II complex on Cape Canaveral to the south and the proposed Nova launch complex to the north.
The diagram shows the distances between the various pads. The Saturn V pads were designed (and LC-39A and LC-39B were built) to be 8,730 feet apart, to limit damage should there be an on-pad explosion. The three Nova pads were planned to be 18,000 feet apart as the Nova rockets were to be much larger.
Wikipedia places Pad 39D immediately west of Pad LC-39C and calls the northern-most emplacement Pad 39E. It being Wikipedia, it cites no sources (or, rather, the source it cites is completely silent regarding the locations of the proposed pads), but since the pads wound up being named from south to north, it would make some sense to have Pad E the farthest north.
As time went on and lunar orbit rendezvous (in which a single rocket launched two, specialized spacecraft, one which was optimized to actually land on the moon and a second spacecraft, the only one which would carry the heavy heatshield required for Earth reentry, would remain in lunar orbit and await the lunar module to rendezvous with it after leaving the lunar surface) was chosen (and budget realities set in), plans for the launch complex were scaled back to meet the projected need.
This (oddly-cropped) diagram shows the results of the initial scale-back of LC-39: It still shows (most of) Pad C but Pad D is depicted with just a dotted line. This diagram is taken from a paper delivered in March 1963, still early enough to realistically expect three pads. The paper notes that "room [is] allowed for future expansion" because
It is difficult, if not impossible, to predict the exact needs for launch facilities past the next few years, but Complex 39 is inherently flexible in regard to additions. Three launch pads are planned for initial construction; however the assembly area for vehicles can service more than that and it can be expanded if necessary. Since so few pads are required at first [because of the reduced time-on-pad resulting from the LC-39 "mobile concept"], real estate is conserved for future pads within a convenient distance of other existing launch support facilities.
So, it is difficult to determine how certain it ever was that Pad D would be constructed or if Pad D was always just part of some "future expansion" allowance, although this crop from a larger Merritt Island Launch Area diagram circa December 1963 has what appears to be the crawlerway for one or more additional pads along with Pads A through C specifically labelled (although with Pad C oddly cropped again):
Another diagram from December 1963 shows the locations of Pads A through C, plus Pad D as a dotted line:
There seem to be a goodly number of diagrams showing LC-39D drawn with a dotted line. This diagram was taken from a document dated February 1966. As SA-500F was rolled out in May 1966, this was far too late to still show to even Pad C (unless perhaps if it were depicted with a dotted line), much less Pad D. In fact, the same document from which I took this diagram also has a similar diagram, but showing LC-39 as built, with only Pad A and Pad B.
The March 1963 paper above also contains a three-pad depiction of LC-39, with no dotted-line version of Pad D:
While LC-39 was still to have three pads when construction began and there are photos of a warning system which included Pad C, Pad C was never to be. However, Pad C did earn the honor of launching the first Saturn V (if only on the the cover of Life magazine):
This undated postcard also shows LC-39 with three pads and a Saturn V launch, although this time Pad B is the launch site (although, with the Spanish galleon just off-shore, it seems more likely that the launch would have been scrubbed or at least delayed):
These discussions of LC-39D, with Pad D being the north-most, do raise some questions. Apparently, standard Missile Test Center practice was to designate the launch pads from north to south, with "A" being the north-most. When construction began on the the current Pad A (i.e., the south-most pad), it was at the time designated "Pad C", in accordance with this practice (the pads were redesignated with Pad A being the south-most pad in January 1963). So, it would seem that plans for Pad D must have been dropped before construction began (since the south-most pad was initially designated "C"), and the south-to-north progression of pads should have gone no farther than Pad C. The appearance of Pad D may simply reflect wishful thinking.
Of course, Apollo 4, the first Saturn V launch, didn't actually occur until November 9, 1967 (NASA didn't quite meet the orange pamphlet's prediction of four launches in 1966 and six in 1967). Other than 1969 (which did see the launch of four Saturn Vs), the launch rate was much lower, with only one or two per year throughout its 1967-1973 operational period. Of the two launch pads at LC-39 actually built, Pad A saw the most use, with Pad B being used only once (for Apollo 10) during the main Apollo program.
After Apollo proper, LC-39 was used to launch Skylab (Pad A used to launch the Orbital Workshop and Pad B to launch the three manned missions) and the Apollo-Soyuz Test Project. The pads were subsequently converted for Shuttle use, with Pad A launching 80 Shuttle missions and Pad B 53. Pad B also launched the Ares I-X rocket, before Barack Obama cancelled Project Constellation.
Post-Shuttle, SpaceX entered into a 20-year lease agreement for Pad A. Pad B will be used with the Space Launch System (SLS), assuming that it survives the whims of the next presidential administration.