Deep Space Navigation: The Apollo VIII Mission · In December 1968 the Apollo VIII mission took three astronauts from the Earth to an orbit around the Moon and back again, safely.

Q U E S T 17:4 20108

F E AT U R E

by Paul Ceruzzi

In December 1968 the Apollo VIIImission took three astronauts from theEarth to an orbit around the Moon andback again, safely. It was the first missionto carry human beings far enough awayfrom the Earth to be influenced primarilyby the gravitational field of another heav-enly body. That mission also marked anumber of other famous milestones: thefirst human crew lofted by the Saturn Vbooster, the Christmas Eve reading fromthe Bible book of Genesis, and thefamous Earthrise photograph showing ablue Earth rising above a barren and for-bidding lunar horizon.

The mission was a triumph of long-distance space navigation. The flight plancalled for the astronauts to arrive at theMoon and establish an orbit that swoopedto an altitude of only about 110 kilome-ters (60 nautical miles, as NASA pre-ferred to measure it) above the lunar sur-face, after a journey of nearly 400,000kilometers (km). They achieved thatnearly perfectly. That was no small feat,as the astronauts were well aware. If theirvelocity was a few percent too low, theywould have crashed into the Moon on itsfar side, out of contact with controllersback on Earth. A few percent too fast, andthey would swing into an erratic orbit,unable to get back to Earth. In an inter-view conducted in 2001 by Steven E.Ambrose and Douglas Brinkley, NeilArmstrong noted that as he and his fellowastronauts were being introduced to thebasics of a mission to the Moon, the nav-igation problem was one of the biggestconcerns that he had:

Well I suppose that everyonewould have concerns, but I don’tknow that they’d all be the same.People would worry about different

things. I remember that one of thethings that I was concerned with at thetime was whether our navigation wassufficiently accurate, that we could, infact, devise a trajectory that would getus around the Moon at the right dis-tance without, say, hitting the Moonon the back side or something likethat, and if we lost communicationwith Earth, for whatever reason, couldwe navigate by ourselves using celes-tial navigation. We thought we could,but these were undemonstratedskills.1

When President John F. Kennedychallenged the nation to send a crew tothe Moon and back before the end of thedecade, navigation was among thebiggest of unknowns. At the beginning ofthe 1960s, it was by no means clear thatnavigating to the Moon would even bepossible. The Pioneer 4 mission,launched in March 1959, indicated thescope of the problem. Its goal was to senda 30 kilogram (kg) probe close enough tothe Moon to measure that body’s radia-tion field, if any. The probe achievedenough velocity to escape Earth orbit, butdue to a timing error in the cutoff of thebooster, it passed by the Moon at a dis-tance of about 27,000 km, beyond therange of its onboard sensors.

One of the first contracts NASA letin preparation for a lunar voyage was tothe MIT (Massachusetts Institute ofTechnology) Instrumentation Labor-atory, under the direction of Charles StarkDraper, for a navigation system. TheInstrumentation Lab had an acknowl-edged expertise in inertial guidance tech-niques, having already developed inertialnavigation systems for aircraft and sub-marines. For space, the lab had donesome preliminary work on a Mars probe,never flown, for the Air Force Ballistic

Missile Division. For a mission likeApollo, inertial navigation had severalchallenges. The first is that gyroscopes,which are at the heart of an inertial sys-tem, tend to drift and give erroneous read-ings. The second is that the system needsto account for the acceleration of gravity,which cannot be distinguished from theaccelerations due to a rocket’s thrust orany other accelerating force.

The drift problem, though serious,was manageable for intercontinental bal-listic missiles (ICBM), whose rocketmotors burn only for a few minutes. Andinsofar as the Earth’s gravity field waswell-mapped, it was possible to factor outgravity for missiles that remained withinthe Earth’s gravitational field. Neitherheld true for Apollo, which had to travelfor several days from Earth to the Moon,and which on arriving at the Moon,entered a gravitational field that waspoorly understood.

Engineers had considered theseproblems and had some experience indealing with them. For the 1957 studydone for the U.S. Air Force, theInstrumentation Lab proposed that theapproximately 150 kg robotic Mars probecarry “a space sextant to make periodicnavigation angle measurements betweenpairs of celestial objects: the Sun, the nearplanets, and selected stars.” Also in the1950s, the Northrop Aircraft Corporationdeveloped a long-range navigation sys-tem for its robotic “SNARK” guided mis-sile, an air-breathing, atmosphericweapon that could navigate across theAtlantic Ocean to targets in the SovietUnion using an automatic star tracker.The Instrumentation Laboratory haddeveloped another inertial system thatoperated for long periods of time, withcorrections for gyroscopic drift. That wasfor ballistic missile submarines, whichremained submerged and hidden as much

Deep Space Navigation: The Apollo VIII Mission

Q U E S T 17:4 20109

as possible, thus precluding periodic sightings on stars, as sur-face ships might navigate. For these submarines, MIT designeda system called SINS: Submarine Inertial Navigation System.Because the SINS gyros would drift over time, the Navy devel-oped the TRANSIT satellite system, which allowed the sub toget a fix on its position by a brief ascent and deployment of anantenna. No optical sighting of stars was necessary. TRANSITwas a radio, not optical system. It was one of the ancestors oftoday’s satellite-based geolocation devices, although its methodof providing a fix, by measuring Doppler shift, did not form thebasis for the Global Positioning System and other modern satel-lite navigation systems.

Space Navigation, before ApolloTo return to Apollo: what else was known about naviga-

tion in space in 1961, at the time the contract with MIT wassigned? The theory of space navigation for piloted spacecraftbegan with the same individual who pioneered in adapting sea-faring navigation for aeronautical use: Philip Van Horn Weems(1889–1979). A graduate of the U.S. Naval Academy in 1912,Weems worked with Charles A. Lindbergh to develop methodsof celestial navigation that Lindbergh and his wife, AnneMorrow Lindbergh, applied in their charting of trans-Pacific airroutes. He is perhaps best known for his innovations in modify-ing chronometers, sextants, and other marine navigation devicesand techniques for use onboard aircraft: innovations that sawwide use among U.S. air forces during World War II.

Weems retired twice from the Navy, but as the space agebegan in the aftermath of Sputnik, he was recalled to active dutyat the rank of captain, and tasked with developing a course onspace navigation for the U. S. Naval Academy.2 The text thataccompanies that course, the Space Navigation Handbook, wasthe work of many collaborators, but its overall tone, and muchof the writing, is probably due to Weems. The course itself, afour-week course at the graduate level, was first convened in thesummer of 1961, a few months after Soviet Union cosmonautYuri Gagarin’s flight marked the dawn of human exploration ofspace. Weems developed aeronautical navigation as an exten-sion of what seamen had done, taking into account the airplane’sthird dimension of altitude, its faster speeds, and other factorsthat precluded a simple extrapolation of existing techniques. Forhis first attempt at developing a theory of space navigation,Weems did the same, starting with what was known about air-craft navigation and extending it into space.

The techniques proposed in that course were quite differ-ent from those proposed by the MIT Instrumentation Lab for theMars probe, or by Northrop for SNARK. Nor were they used forApollo missions, even though they were centered on the astro-naut’s making observations of Earth, Moon, and stars from thespacecraft. One principle he developed was to determineabsolute distance from the Earth (or Moon) by measuring thesize of the observed disk, and to further determine the space-craft’s position in space by observing the star field behind theEarth of the Moon during such an observation. For trajectoriesclose to the Earth, the spacecraft’s altitude could be inferred byobserving the amount of Earth’s curvature. No inertial deviceswere proposed. Some amount of onboard computation was

required, but Weems proposed, as he had done successfully forair navigation, to precompute solutions in advance and providethe astronaut with that data as printed tables or graphs. Weemswas reluctant to employ an onboard electronic computer as atthat time, circa 1961, such devices were neither reliable norcompact. (The SNARK guidance system, mentioned above, didhave onboard computational ability, though it did not carry adigital computer. The device used vacuum tubes.)

Among the navigational aids mentioned by Weems was anelectromechanical “Position Finder” designed by Edwin Collenof the Kearfott Division of General Precision. (Collen called itan “Astronavigator,” and that term will be used in this essay.)The device contained a small globe, a transparent hemisphericalstar chart that depicted stars as dots of luminescent paint, and aningenious system of lenses and mirrors that allowed the astro-naut to superimpose his observations of Earth and star field withthose of the small globe and painted star field. When the twowere aligned, the astronaut could determine a position in spaceand altitude above Earth. Collen proposed that Apollo astronautscarry such a device as an emergency backup to the main Apolloguidance and navigation system. NASA rejected the proposal,although it did develop a set of emergency procedures, some ofwhich were used during the Apollo XIII mission.3 For the earlySoyuz missions, Soviet cosmonauts carried a similar deviceonboard to help them locate their position in orbit.4

For Project Mercury, NASA adopted an even simplermechanical navigator, intended to assist the astronaut in return-ing to Earth. It consisted of a small globe rotated by mechanicalclockwork, with an icon depicting the Mercury capsule suspend-ed above it. On achieving a stable orbit, the astronaut would setthe capsule’s inclination and period of the orbit, as radioed upfrom the ground. The globe rotated under the icon to follow therotation of the Earth, and another mechanism adjusted the orbitto account for the precession of the capsule’s orbital plane. Thusthe astronaut could follow a path as the icon passed over the

“Earth Path Indicator,” Project Mercury. Credit: National Air & Space Museum

Q U E S T 17:4 201010

globe. John Glenn carried this “EarthPath Indicator” on the Friendship 7 cap-sule in February 1962, but during thatflight Glenn’s naked-eye observations ofthe Earth were so good that NASA foundthe device redundant. It was carriedonboard the following Mercury flight,piloted by M. Scott Carpenter in May1962, and not used again.5

These examples of early spacenavigation devices illustrate an importantpoint: the early 1960s was a time of enor-mous advances in the science and tech-nology of spaceflight, including naviga-tion. In those few years, engineersimproved gyroscope technology andreduced drift. The mathematics of celes-tial mechanics, already well developedsince the time of Newton, was directedtoward practical spacecraft mission plan-ning. Computers made a transition fromvacuum tubes to solid state and becamemore powerful and more reliable.Ground-based computers remainedlarge, but the development of high-levelprogramming languages like FORTRANmade them more capable of handlingcomplex mathematical problems. Large-diameter radio antennas were placed intoservice and provided an ability to trackspacecraft into deep space. Atomicclocks were developed that were manytimes more accurate than the quartz-

based clocks in use in the 1950s. Finally,with each Mercury and Gemini flight,NASA gained valuable data on the per-formance of navigation techniques byreal-world experience. This was truethroughout the decade but was especiallypronounced as the Gemini program tran-sitioned to Apollo.

With President Kennedy’s end ofthe decade goal rapidly approaching,Apollo engineers had to make decisionsabout the craft’s design and “freeze” iteven as, for example, the Gemini XI mis-sion, in September 1966, set an altituderecord by coupling a Gemini spacecraftto an Agena and boosting its orbit to theedges of cislunar space. Thus, althoughthe Gemini inertial platform used fourgimbals, Apollo used only three, a sim-pler system that however could lead to acondition known as “gimbal lock.”6

Under normal conditions the crew couldavoid gimbal lock, although in the ApolloXIII mission, gimbal lock was a persist-ent and recurring threat. Note that astro-naut James Lovell, who flew on bothApollo VIII and XIII, was also a veteranof two Gemini missions: Gemini VII andXII; thus he had extensive experiencewith both designs, and clearly favoredthe four-gimbal system. In mostinstances, however, Apollo was able totake advantage of advances in technolo-

gy made earlier in that decade, mostnotably Apollo’s use of integrated cir-cuits in its guidance computer. But insome cases the Apollo VIII mission car-ried with it some of the legacy of deci-sions made earlier in the decade based onconditions that had changed.

The MIT Instrumentation Labora-tory’s Approach

Weems’s work laid a foundationfor thinking about navigation in space ata time when only a few steps had beentaken into that realm. By the late summerof 1961, when NASA awarded the firstcontract to the MIT Instrumentation Labfor Apollo navigation and control, a dif-ferent approach to navigation emerged.7Central to Weems’s approach was thedetermination of one’s distance from theEarth by measuring the angle of theEarth’s disc as seen from the spacecraft.The Instrumentation Lab’s approach wasmore traditional: to measure the anglebetween the Earth’s horizon and a star orbetween two stars.8 These techniqueswere initially developed under an AirForce contract for a robotic mission toMars, never flown. After 1958, the newcivilian agency NASA took over plan-ning for deep space missions from theAir Force. In November 1959, NASAdirected the Instrumentation Lab to workwith the Jet Propulsion Laboratory inPasadena, California, to plan deep space,robotic missions. But according to DavidHoag of the Instrumentation Lab, the twolaboratories diverged in their approachesto navigation; with JPL focusing less ononboard techniques and more on navigat-ing by precise radiometric techniquesfrom the ground.9 This divergence wouldreappear during the human Apollo mis-sions out of Earth orbit, as we shall see.

The approach adopted by theInstrumentation Laboratory is bestdescribed as an inertial system, periodi-cally corrected by star sightings. Theinertial component was based on theLab’s work on the guidance system forthe Polaris submarine-launched ballisticmissile, which did not need stellar cor-rection, as it was designed to operate fora brief period of time. For Apollo, theposition given by the inertial systemwould be corrected, perhaps twice a day,

SNARK guidance system, with automatic star-tracker. Credit: National Air & Space Museum

Q U E S T 17:4 201011

by sightings on the Earth, Moon, or thestars. For that purpose Apollo carried asextant, manufactured by the KollsmanInstrument Corporation, which operatedlike a traditional maritime sextant in thatthe astronaut would adjust the deviceuntil the image of a known star wasaligned with the image of, say, the Earth’sor Moon’s horizon. That angular datumwas then fed into the Apollo GuidanceComputer (AGC), which computed thespacecraft’s position. The sextant did notlook like a classic marine or even aircraftsextant, but that term was appropriate. Ithad two telescopes: one at 28 power andthe other at unity power but with a widerfield of view. The optics were alsomechanically constrained in their move-ment because of the need for the tele-scopes to penetrate the pressure hull.10

The Instrumentation Lab devel-oped a basic technique to correct for drift.The astronauts would key in the codednumber of a given star, and the computerwould then orient the spacecraft’s opticsso that the star was centered in the eye-piece of the telescope. Depending onhow much the inertial system had drifted,the star would be found a short distanceaway from the crosshairs. The astronautwould center the image of the star, pressa button, and the computer would notethe amount of drift. The process could berepeated against a second star if neededto realign the gyros.

Thus, although the Apollo naviga-tion system was far removed from whatWeems had envisioned a few years earli-er, it replicated nautical techniques in usefor centuries. It used a sextant. Bothsailors and Apollo astronauts carried starcharts. Sailors carried books of mathe-matical tables to assist them in convert-ing observations into latitude and longi-tude; Apollo astronauts carried similarmathematical data, in this case precom-puted and stored in the onboard comput-er. One should not push the analogy toofar. Apollo was fundamentally an inertial,not a celestial system. Still, the notion ofdeep space as “this new ocean,” inPresident Kennedy’s words, must havebeen strong.11

Ground-Based TrackingAs the 1960s progressed, the end-

of-the-decade deadline imposed by

President Kennedy grew moreurgent. Like many of the sys-tems that made up Apollo, thedevelopment of the guidancecomputer was not easy. In par-ticular, the software being writ-ten for the missions kept grow-ing, overwhelming the memo-ry capacity of the computer.Many of the accounts of thedevelopment of the systemfocus on the hardware, correct-ly emphasizing the break-through of using the newlydeveloped integrated circuit,and the heroic efforts made toensure reliability, while avoid-ing the complexity and weightof having redundant hardware(as the IBM-designed LaunchVehicle Digital Computer onthe Saturn V had).12 It turned out that apacing element was the writing and test-ing of the software, especially as theAGC was tasked not only with naviga-tion but also with controlling the servicemodule rocket motor engines digitally(the Saturn V’s rockets were controlledby a separate, analog computer).

The crisis came to a head in May1966, at a meeting held at MIT andchaired by Howard W. (“Bill”) Tindall ofNASA.13 As a result of this and otherfollowing meetings, a number of tasksfor the AGC were eliminated, to savememory. And the question of providing aredundancy in the event of a computerfailure was resolved by designatingground-based navigation techniques asthe back-up to the onboard computer.This finessed the on-going debate aboutwhether to have the computer repairableinflight by the crew, who would be pro-vided with spare modules that they couldinstall in the machine. Prior experiences,especially Gordon Cooper’s MercuryMA-9 flight of May 1963, indicated thatmaking repairs to onboard electronicequipment created as many problems asit might solve.14 Apollo would fly withonly one AGC, sealed from the elements,and designed to work as reliably as pos-sible. MIT engineers’ memoirs of theApollo missions stress this fact: the AGChad to work right the first time, and it did,throughout the program.15

Regardless of that decision, theneed for onboard navigation abilityremained, at least to navigate if commu-nications with Earth-based stations werelost or degraded. There were also twomajor components of the Apollo mis-sions that had to be controlled onboard.The first was to reduce the velocity of theApollo configuration to enter into a sta-ble orbit around the Moon. The physicsof orbital transfer dictated that this beperformed by a carefully controlled burnof the service module’s engine behind theMoon, out of contact with Earth. As men-tioned above, the resulting orbit was tohave a pericynthion (the lunar equivalentof perigee for Earth orbit) of approxi-mately 60 nautical miles, so this maneu-ver had to be executed correctly. The sec-ond function was the landing itself. Thetime delay of radio signals from Earth tothe Moon is not much, on the order of afew seconds, but that was too long toentertain any notion of controlling thelanding from Earth. Strictly speaking,this second function was a control, not anavigation problem, but the absoluterequirement for onboard guidance andcontrol capability meant that onboardnavigation ability had to be present.

Although the AGC did not useredundant circuits, as the Saturn VLaunch Vehicle Digital Computer used,

Collen “Astronavigator” Credit: National Air & Space Museum

Q U E S T 17:4 201012

there were redundancies onboard. Thelunar module (LM) carried an identicalAGC, which could be used as a backupon the outward leg of the journey. It wasunavailable for the return to Earth, as theLM ascent stage was jettisoned beforethat leg of the journey. (Note also that theApollo VIII mission carried no LM oneither leg of its mission.) The LM alsocarried an Abort Guidance System(AGS), intended to be used only to getthe LM’s ascent stage off the Moon andinto lunar orbit quickly in the event of anemergency. The AGS contained a smalldigital computer of its own.16 NASArejected electromechanical backupdevices, such as Edwin Collen’sAstronavigator, but astronauts could nav-igate using sightings out the windows,with timing supplied by their mechanicalanalog wristwatches. Nearly all of theseprocedures, including the use of the LM’sguidance computer, were used to bringthe Apollo XIII astronauts home safely. 17

The ultimate outcome of the meet-ings between NASA and theInstrumentation Laboratory was thatground-based navigation techniqueswould play a greater role.18 By 1966,with the transition from Gemini to Apollofully underway, the policy went beyondthat: “The primary navigation system incislunar space is the ground system[emphasis added].”19 By that year,NASA had established a world-circlingnetwork of nine-meter (30-foot) antennasfor tracking, communications, and con-trol at sites around the globe. For Apollo,

these sites were used for the Earth-orbitalphase of the mission. To track the astro-nauts after leaving Earth orbit, only threesites with larger, 26-meter antennas,spaced approximately 120° longitudeapart, were needed, although the 30-footstations were also used. These were co-located near the three Deep SpaceNetwork (DSN) sites managed by theJPL for unpiloted missions away fromEarth orbit.20 The DSN and the MannedSpace Flight Network (MSFN) had dif-ferent missions but were complementary:each used 26-meter (85-foot) antennas,and links were established so that theDSN system could back up MSFN if nec-essary. The three stations were atGoldstone, California; HoneysuckleCreek, near Canberra, Australia (about 30km away from the DSN Tidbinbilla sta-tion); and Fresnedillas, Spain, west ofMadrid.21

The MSFN did not duplicate themuch larger DSN 70-meter (230-foot)antennas, which the DSN required for tra-jectories far beyond the orbit of theMoon. The links between the two sys-tems, however, allowed NASA to usethem, in addition to large radio astrono-my dishes, such as the 64-meter (210-foot) antenna at Parkes, Australia, ifneeded—a need that did arise on severaloccasions.22

The decision to use this network asthe primary cislunar navigation systemwas because of more than just the limita-tions of the onboard AGC. One factorwas political: by the mid 1960s, the fearthat the Soviet Union would jam or other-wise interfere with communicationsbetween astronauts and the ground duringa lunar mission abated. The space racewas still on, and classified informationabout Soviet progress was communicatedat least to James Webb, NASA adminis-trator, at the time. The decade began witha sense that space would be a militarytheater much like air: with orbitingbombers poised to deliver bombs to tar-gets on Earth, with interceptors shootingenemy spacecraft, and so on. That did notmaterialize. ICBMs travel through spaceon their way to a target, and space didbecome a theater for military reconnais-sance and signals intelligence, but formilitary activities the human presencewas found to be unnecessary, impractical,

or not worth the cost. Although tensionsremained high between the United Statesand the Soviet Union during the 1960s, asthe decade progressed NASA came tobelieve that that the Soviet Union was notgoing to directly threaten U.S. humanmissions to the Moon, and the fear ofjamming subsided.

Among the technical advancesbehind the decision was the developmentfor Apollo of a unified system that com-bined all tracking, communications,uplinked commands, telemetry, and othercontrol signals onto one band of frequen-cies in the S-Band, around 2,300 MHz.23

This system was not in full operation atthe time of the Apollo VIII mission, butenough was in use to allow more func-tions to be performed by Apollo’s avion-ics without incurring a weight penalty orexcessively complicating the spacecraft’sdesign. A second technical factor was therapid improvement in the accuracy oftiming devices, beginning with the intro-duction of rubidium and cesium-basedfrequency standards in the late 1950s,which replaced quartz oscillators that hadbeen in use since before World War II. Bythe late-1960s, cesium-based clocksattained a stability of nearly 1:1013

defined as the stability of the standardduring a 24-hour period. That was nearly100,000 times improved over quartzoscillators (and a million times betterthan mechanical chronometers).24 Forthe Apollo missions, NASA’s MSFNused a rubidium standard, not as accurateas cesium but stable enough to give veryaccurate fixes.25 The best fixes wereobtained using the 26-meter antennas onthe ground, and the high-gain, S-bandantenna located on the Apollo servicemodule, with its distinctive array of foursmall parabolic dishes pointed towardEarth. A fix could also be obtained usingthe lower-power, omnidirectional anten-na on the spacecraft, thus providing anadditional layer of redundancy.26

The early 1960s also was a time ofadvances in radio communications andsignal processing, which further con-tributed to meeting President Kennedy’schallenge. Around the time of Sputnik,Eberhardt Rechtin of JPL was workingon jamming, and defense from jamming,of signals for guided missiles. In thecourse of that work he turned to the clas-

A Soviet Soyuz navigatorCredit: National Air & Space Museum

Q U E S T 17:4 201013

sic mathematical work done by NorbertWiener during World War II on theextraction of a meaningful signal in thepresence of noise. Using that as a startingpoint, Rechtin and his colleagues cameup with techniques that would allow oneto track spacecraft with great accuracyfrom the ground, despite the weight andsize restrictions that made it difficult forspacecraft to carry high-powered trans-mitters or large parabolic dishes onboard.One of the outcomes was the develop-ment of the Phase Lock Loop (PLL),which is now the standard method ofreception employed by cell phones, GPSreceivers, car radios, broadcast and satel-lite televisions, et cetera. In an interviewconducted for the IEEE (Institute ofElectrical and Electronics Engineers) in1995, Rechtin recounted how experts toldhim that it would be impossible to receivemeaningful information from deep space,as the weak signals would be swampedby background noise. The PLL used anarrow band receiver, which tracked thefrequency of the transmitter even as thatfrequency shifted due to Dopplereffects.27 Rechtin was also among thosewho developed the concept of using asequence of “pseudo-random” num-bers—a sequence of digits that appearedto be random but that were known andspecified in advance, to carry informa-tion. This, too, was an outgrowth ofNorbert Wiener’s work on the extractionof signals from noise, and would bedeveloped by others into what hasbecome known as “spread-spectrum”communications in common use today.

The ground-based navigation sys-tem used for Apollo employed these tech-niques. A pseudo-random code was sentto the spacecraft over the unified S-band.It was received, and retransmitted back toEarth on a slightly different frequencyalso within the S-band.28 The time ittook for the signal to go to and from thespacecraft, subtracting out the knowntimes for the signals to travel within theequipment, divided by two, gave theabsolute distance of the spacecraft fromthe dish, to within 2 meters.29 The angleof the antenna as it focused on the space-craft gave further information on thecraft’s position, to a few milliradians(about 1/10°). Finally, a measurement of

the Doppler shift of the S-band frequencygave the radial component of the space-craft’s velocity relative to the ground sta-tion. By combining readings from differ-ent ground stations, plus similar readingstaken during a span of time, the space-craft’s velocity and position could bedetermined in all axes.

It was the precision of the Dopplermeasurements that tipped the balance infavor of ground techniques. By an ingen-ious use of the telemetry codes, com-bined with using more than one groundstation to track the spacecraft and therubidium frequency standard, NASA wasable to determine range to around 30meters, and velocity to within 0.2 feet persecond.

Simulations done before the ApolloVIII mission showed that tracking by theground-based MSFN was more accurateduring the initial phase of a mission, untilabout 35 hours after Trans LunarInjection (TLI), when the spacecraft wasabout halfway between Earth and theMoon. From that point until the craftentered lunar orbit, the onboard systemwas slightly more accurate. MSFN-basedtracking was also more accurate in deter-mining position and velocity as thespacecraft orbited the Moon. That was

because of the geometry of the Apollo’strajectory as it orbited the Moon; perhapsalso to the irregularities of the Moon’sgravitational field.30

These techniques determined theposition of spacecraft beyond Earth’sorbit with great accuracy. Two decadeslater, the Global Positioning System(GPS) would use the same techniques“inside-out” to determine the position ofa person or object on Earth. GPS usespseudo-random codes, and measures dis-tance by measuring the time it takes a sig-nal to get to a receiver from space. Butthe GPS system has the atomic clocks inspace, not on the ground. And GPSknows the position of the satellites inspace, not the receiver on the ground, togreat accuracy.

Many histories of the Apollo pro-gram have focused on the remarkablecapabilities of the AGC, with its novelinteractive programming ability, its relia-bility, and its pioneering use of integratedcircuits. Fewer have looked at the com-puters used on the ground to support themission. NASA’s computing facilitiesgrew out of a Naval Research Laboratoryfacility on Pennsylvania Avenue inWashington, within sight of the Capitol.With the establishment of NASA in 1958,

James Lovell, sighting through the Apollo sextant, Apollo VIII. His right hand hovers overa button, which marks the sextant’s position as it is centered over a known star. On theupper right is the Apollo Guidance Computer’s Display Keyboard (DSKY), one of two inthe Command Module. Credit: NASA

the Space Computing Center moved toGreenbelt, Maryland, at the GoddardSpace Flight Center. Beginning in 1960,Goddard computers, primarily IBM(International Business Machines) main-frames, calculated trajectories and orbitsfor robotic and early human flights.NASA was one of IBM’s best customers,and it was able to do what few other IBMusers could do, namely make fundamen-tal modifications to the systems IBMsupplied. IBM leased, not sold its com-puters, and it did not allow its customersthat freedom. In a typical mainframeinstallation, programs were keypunched,transferred from decks of punched cardsto tape, and the tapes ran through thecomputer to give an answer in printedform. IBM made an exception to that rulefor NASA, which needed to be able torun the computers in “real-time”: to enterin data directly and receive results assoon as the computer could calculatethem. In other words, NASA modifiedthe IBM machines to operate like a mod-ern personal computer, even if themachine cost a million dollars andrequired an air-conditioned room of itsown .

For Project Mercury, NASA creat-ed a “Mercury Monitor” system that

could operate this way. By the mid-1960s, NASA–Goddard continued tomanage space communications, whilenavigation and trajectory analysis weretransferred to the Real Time ComputationCenter (RTCC) at the MannedSpaceflight Center (MSC) in Houston.Initially the RTCC used a set of IBM7090-series mainframes, which werereplaced beginning in 1966 by IBM’sthird-generation System/360 comput-ers.31 The term “third generation”implied that they used integrated circuits(IC). The System/360 used a hybrid cir-cuit called “Solid Logic Technology,”which combined a number of discretecomponents onto a small ceramic sub-strate. (The AGC circuits used silicon andwere a direct ancestor of the ICs in usetoday). The new-generation IBM main-frames were a large advance in the stateof computing art. They not only hadfaster processing speeds and more mem-ory, they also came with more sophisti-cated software. Programmers at MSCdeveloped a customized operating sys-tem that allowed Houston controllers tooperate the computers in both batch andreal-time mode. Called HASP (HoustonAutomatic Spooling Priority; SPOOLwas itself an acronym related to

input/output functions), it not only servedNASA well but also was offered as aproduct by IBM to other customers.

Creating an operating system for amachine as complex as the System/360was perhaps one of the most challengingtasks in all of computer programming.The fact that NASA was able to do this sowell, while its main focus was gettinghuman beings to the Moon, is testimonyto the space agency’s talent.32 These sys-tems created an effervescent atmospherein Houston, as mathematicians combinedcenturies-old equations of celestialmechanics developed by Isaac Newton,Pierre-Simon Laplace, Carl FriedrichGauss, and others, with new techniquestailored for space missions and taking fulladvantage of the number-crunching abil-ity of IBM’s hardware.

Apollo VIIIThese concepts and simulations

were put to the test on the morning of 21December 1968, with the launch of aSaturn V rocket, carrying a crew of threeastronauts, from Pad 39A of the KennedySpace Center in Florida. About 38 min-utes after launch, and while still in low-Earth orbit, command module PilotJames A. Lovell Jr. jettisoned the protec-

Q U E S T 17:4 201014

Major units of the CM Guidance,Navigation, and Control System.Note the two DSKYs: one on themain panel, the other next to thescanning telescope and sextant. Credit: Charles Stark DraperLaboratories

Q U E S T 17:4 201015

tive covers from the command moduleoptics, in preparation for a preliminaryalignment of the spacecraft’s inertial plat-form.33 When he looked through the tel-escope, he saw a field of bright particlesthat made it hard to locate stars. The par-ticles were probably small pieces ofdebris that came off the spacecraft whenthe covers were ejected. Later on duringthe mission there would be other sourcesof highly-reflective particles that wouldcomplicate the procedure. The crew alsofound it necessary to dim the cabin lightsso that Lovell’s eyes could identifyobjects in space. It took some time, about15 minutes, to sort out the procedures, butLovell was able to locate two stars, centerthem in the optics, and command theonboard computer to realign the gyros.34

Lovell’s readings, plus the computer’scalculations, aligned the gyros to within.01° of what ground-based tracking indi-cated. “Pretty good for a beginner here,”remarked lunar module Pilot William A.Anders.35

The real test of the system camelater, after the crew left Earth orbit. Afterthe launch and subsequent injection intoEarth orbit were determined to be suc-cessful, Michael Collins, from his con-sole at Mission Control in Houston, gavethe crew the go ahead to restart theSaturn’s third stage and send the astro-nauts to the Moon. Collins gave the mes-sage “Apollo VIII, you are go for TLI[trans-lunar injection]” at about twohours and 27 minutes after launch. At twohours, 50 minutes, the third stage of theSaturn V fired again for about five min-utes, which increased its velocity, and thecommand and service modules attachedto it, enough to reach the vicinity of theMoon.36

As might be expected for ApolloVIII, the crew and mission controllers inHouston had to adapt and modify theirplans as the mission proceeded. The TLIburn was executed perfectly, as was theseparation of the Apollo command andservice modules from the Saturn V thirdstage. The schedule then called for the

crew to take readings on stars and checkthe Inertial Measurement Unit (IMU)alignment shortly after separation, but thecrew found that the third stage was flyingin formation with them a little closer thanexpected, and they spent a bit of timeensuring that there would not be a colli-sion. That put them behind schedule, butwith the Moon several days away, it wasnot a critical delay. At about five hoursinto the mission, ground controllersdirected the third stage to dump its resid-ual propellants and other fluids, to pre-vent a possible explosion. That had theadditional effect of pushing the stageaway, although that had to be carefullychoreographed. The procedure, however,dumped a lot of small particles into theregion around it, and as the Sun reflectedoff them, the astronauts once again hadtrouble distinguishing between the parti-cles and stars.

Partly for that reason, a procedureto calibrate the Apollo optics wasscrapped, and the crew proceeded direct-ly to take sightings and feed that informa-tion to the IMU.37 The resulting readingswere way off what the ground trackinghad indicated. Mission controllers decid-

ed to ignore the readings. At this point asecond complication arose: the missionplan called for the crew to rotate the com-mand service modules (CSM) slowly, toeven out the effect of solar radiation(called “Passive Thermal Control,” betterknown as “barbecue mode”). That had tobe postponed, as it would conflict withthe need to hold the CSM to a precise atti-tude to locate a star. 38

Eventually all of this was workedout. The crew managed to remove thebias from the optics, find stars, and takegood readings. Working with ground con-trollers, they learned how to roll thespacecraft for Passive Thermal Control,and stop it when necessary to point thetelescope at a target. By about 12 hoursinto the mission, the onboard navigationcalculations, carried out by the AGC,were in close agreement with the ground-tracking data. All indications showed thatwhen the astronauts arrived at the Moon,they would skim above its surface--butnot hit it—and thus be able to fire theservice module’s engine to put them intoa safe and stable orbit around the Moon.

As the crew got closer to the Moon,their navigation readings got progressive-

Apollo Guidance Computer DSKY on themain panel, during a ground simulation.

Credit: NASA

Q U E S T 17:4 201016

ly easier. Early in the journey, sunlightreflected from the Earth washed out someof the sightings through the telescope;this diminished as the Earth receded. Thespacecraft continued to be surrounded bystray particles, including ice crystals aftera urine dump, but it became easier to dis-tinguish between these and the stars. Alsoas the Earth receded and as Lovell gainedpractice, he was more consistent in sight-ing the Earth’s horizon, which was indis-tinct because of its atmosphere. A readingof the angle between the Earth’s horizonand a star, taken at 17 hours, 53 minutes,into the mission, matched perfectly withwhat the onboard computer indicated.Lovell exclaimed, “How’s that, sportsfans? All balls.” To which Anders replied,“As soon as you’ve got an audience, youdo great.” (“All balls” was the astronaut’sway of saying the computer DSKY—data storage and keyboard—was display-ing all zeroes, that is, no error.)39 On thereturn journey, the onboard MIT naviga-tion system continued to work perfectly.The mid-course corrections were mini-mal, and the onboard navigation readingsmatched those from MSFN exactly. Nearthe end of the mission, around the time ofthe final mid-course correction,Astronaut Jerry Carr, on duty as capsulecommunicator at Mission Control, jokedthat Lovell’s head was getting swollenbecause of the good job he was doing: “Ihate to tell you this Frank [Borman],

because Jim probably won’t even be ableto wear his comm. carrier [his communi-cations headset] anymore, but that last setof marks put your state vector right on topof the MSFN state vector.”40 This banterbetween the crew and Mission Controlreflected the fact that the Apollo VIII mis-sion, which began with so manyunknowns and risks, was turning out tobe a huge success. It met all its objectivesand gave NASA confidence thatPresident Kennedy’s goal to land a manon the Moon would in fact be met.

However, there is more to thisstory, and to understand that we go backto an exchange between the crew andMission Control at about nine hours intothe mission, as Apollo VIII was preparingfor its first mid-course correction. Thespacecraft was on a very good course, butmission planners decided to allow itdeliberately to drift a little farther thannecessary before making a correction, sothat the mid-course correction could bemade with the service module’s SPS(service propulsion system) engine,rather than the small thrusters that wouldhave sufficed for a small correction. Thisallowed NASA to test the engine, whichhad to perform a critical maneuverbehind the Moon twice: once to getApollo VIII into lunar orbit, and again tobring it home. Because Apollo VIII wasnot carrying a lunar module, there was noredundancy; the SPS engine had to

work.41

This test of the SPS engine wascritical, on which the success of the mis-sion, and with it President Kennedy’sgoal, depended. Reading between thelines of the conversations between thecrew and Mission Control, one senses thetension that everyone was feeling. Therewas none of the good-natured banter asthe crew prepared for this burn. Beforethe mid-course correction, it was neces-sary to get as accurate a fix on the space-craft’s position as possible, and the crewmethodically took readings and enteredcommands into the computer. Theyfound, to their relief, that the resultingnavigational fix from the onboard systemwas in close agreement with the statevector computed on the ground. Nowcame a critical decision for mid-coursecorrection: which to use?

NASA decided that the trackingfrom the ground would take precedence.At about 10 hours into the mission, andan hour before the critical mid-coursecorrection, Frank Borman asked MissionControl if it was time to realign theonboard platform. From his console inHouston, Capsule CommunicatorMattingly said, “That’s negative, ApolloVIII. We would like to update things first,and we’re going to give you an LM statevector and then an external Delta-V.”42

What that meant was that NASA wouldtransmit the craft’s state vector—the setof numbers describing its position andvelocity—from ground-based, or exter-nal, data taken from its tracking network,and store that into a portion of the AGC’smemory location that was availablebecause there was no LM on this mission.That state vector would be used as thebasis to compute the mid-course correc-tion. Borman keyed in the “Verb” P00into the computer, which essentially toldit to suspend all other program execution.He then flipped a switch, located just to

Comparison of ground-based (MSFN) andon-board (AGC) navigation accuracy at var-ious phases of a mission, projected in sim-ulations, 1966. (Ref: Proceedings, ApolloLunar Landing Symposium, June 25-27,1966, Houston, Texas, NASA, NASA-S-65-10006; fig. 24. Credit: NASA

Q U E S T 17:4 201017

the left of the navigation station’s DSKY,from “Block” to “Accept”: allowing thecomputer to receive and store the databeing sent from Earth.

For the rest of the Apollo VIII mis-sion, and for all Apollo missions there-after, ground navigation data took priori-ty for all navigation maneuvers outsidelow Earth orbit. Ground navigation tookpriority on all subsequent missions. It wasespecially critical during the Apollo XIIImission of April 1970, when an explosioncut off power to the CM, leading to thepowering down of the CM’s guidancecomputer. The crew—which coinciden-tally included James Lovell—used theLM computer for guidance, but not fornavigation. Using a combination ofground commands, plus onboardmechanical aids and mechanical wrist-watches, they were able to safely reenterthe Earth’s atmosphere and land within afew kilometers of the intended landingpoint.43

The Apollo VIII astronauts realizedthe dream of autonomous human spacetravel, in which spacefarers followed inthe footsteps of Captain Cook, or theLewis and Clark expedition.44 Althoughthey did not land on the Moon, theiraccomplishment will stand for all time asone of the greatest in the annals of humanexploration. But we must also rememberthat the dream lasted only about sevenhours: from the time the crew was givena “Go for TLI,” at three hours into themission, to flipping the switch to“Accept” at ten hours into the mission.

As of this writing, no human

beings have had to navigate outside theclose range of Earth since Apollo XVII in1972, but one may assume that such mis-sions will resume eventually. When theydo, how will the crew navigate? Since1972 we have seen remarkable achieve-ments with unpiloted deep space probeshitting very precise targets at the moonsof outer planets, asteroids, and comets,beginning with the Mariner 10 mission toVenus and Mercury in 1974. But we mayalso assume that there have beenadvances in onboard, stellar-inertial tech-niques also. Whether the dream ofautonomous space exploration can berevived remains to be seen .

About the AuthorPaul E. Ceruzzi is curator of aerospaceelectronics and computing at the NationalAir and Space Museum in Washington,DC. Dr. Ceruzzi attended Yale Universityand the University of Kansas, from whichhe received a PhD in American studies in1981. He is the author or co-author ofseveral books on the history of computingand related topics.

Notes1. Neil A. Armstrong, interview with Dr.Stephen E. Ambrose and Dr. DouglasBrinkley, Quest 10, no. 1 (2003): 6–45;quote on 29–30.

2. [U.S. Navy], Space Navigation Handbook,NAVPERS 92988 (Washington, DC: U.S.Government Printing Office, 1962), i.

3. Edwin Collen, private communication tothe author. Collen donated the prototypeAstronavigator to the National Air and Space

Museum, catalog number A 1993-0078-000.

4. One of the devices, used by cosmonautVladimir Shatalov and carried on Soyuz 4 in1969, is on display at the National Air andSpace Museum.

5. During his tenure in the Senate, Glenndisplayed an Earth Path Indicator (EPI) on acredenza behind his desk. The EPI in theNational Air and Space Museum collections,catalog number A 1972-1170-000, wastaken from an earlier, robotic Mercury flight.

6. David Hoag, “The History of Apollo On-board Guidance, Navigation, and Control,”Ernst A. Steinhoff, editor, The Eagle HasReturned (American Astronautical Society,supplement to Advances in theAstronautical Sciences, vol. 43, 1976), 277.

7. David A. Mindell, Digital Apollo: Humanand Machine in Spaceflight (Cambridge,Massachusetts: MIT Press 2008), 106–107.

8. Hoag, “The History of Apollo,” 270–300.

9. Hoag, “The History of Apollo,” 271.

10. Although it was not difficult to maneuverthe stack while coasting, during an observa-tion the crew had to suspend the so-calledPassive Thermal Control, or “barbequemode”: the slow rotation of the craft to evenout the heating of solar radiation.

11. President Kennedy used this phrase at aspeech at Rice University in Houston on 12September 1962. Although not the casewith Apollo, note that the Space Shuttleorbiters were named after ships of discov-ery. The Shuttle Enterprise was named aftera fictional spacecraft from the televisionseries Star Trek. But that Enterprise wasitself named after a ship.

12. Eldon C. Hall, Journey to the Moon: The

Apollo 11 star charts. This was prepared especially for the position of the stars, Earth, Sun, and Moon at the time of the Apollo 11 launch,July 1969. Slightly different charts were prepared for Apollo VIII and would have been carried onboard. Note the two-digit numbers formany of the stars—these numbers were keyed in as a “noun” into the Apollo Guidance Computer’s DSKY. Credit: NASA

History of the Apollo Guidance Program,AIAA, 1996 13. Hoag, “The History of Apollo,” 290.

14. Chris Kraft, Flight: My Life in MissionControl (New York, Dutton, 2001),182–183.

15. Hoag, “The History of Apollo,” 284.

16. James Tomayko, Computers inSpaceflight: The NASA Experience (NewYork: Marcel Dekker, 1987), 59–60.

17. John L. Goodman, “Apollo 13 Guidance,Navigation, and Control Challenges,”Proceedings AIAA Space 2009 Conferenceand Exposition, Pasadena, California,September 2009.

18. “More Apollo Guidance FlexibilitySought,” Aviation Week and SpaceTechnology (16 November 1964): 71–72.

19. Robert C. Duncan, “Apollo Navigation,Guidance and Control,” Houston, Texas:NASA, Manned Spaceflight Center,Proceedings and Compilation of Papers,Apollo Lunar Landing Symposium, 25-27June 1966, unpaginated. Emphasis in theoriginal.

20. Sunny Tsiao, “Read You Loud andClear!”: The Story of NASA’s SpaceflightTracking and Data Network, (Washington,DC: NASA, 2008), NASA SP-2007-4232,146.

21. Tsiao, “Read You Loud and Clear!”,146–147.

22. Live television during the Apollo 11landing was received by the large antennasin California and Australia.

23. Myron Kayton, editor, Navigation: Land,Sea, Air, and Space (New York: IEEE Press,1990), 317–331.

24. William G. Melbourne, “Navigationbetween the Planets,” Kayton, editor,Navigation, 337. The author notes that proj-ect Mercury kept time using the NationalBureau of Standards radio station WWV,accurate to about one millisecond.

25. Rubidium clocks were estimated tokeep time to within one second in 30 years,about 1,000 times less accurate thancesium.

26. Duncan, “Apollo Navigation, Guidance,and Control,” Proceedings. During theApollo XIII mission, the hi-gain antenna,located on the damaged service module,was unusable. The astronauts communicat-ed with low-power transmitters located onthe command and lunar modules, throughthe 70-meter DSN antennae pressed intoemergency service.

27. Eberhardt Rechtin, interview withFrederik Nebeker, IEEE Center for theHistory of Electrical Engineering, 23February 1995, accessed at

h t t p : / / w w w . i e e e g h n . o r g / w i k i /i n d e x . p h p / O r a l - H i s t o r y :Eberhardt_Rechtin.html. Reprinted in Quest15, no. 2 (2008): 20–39.

28. The downlink carrier frequency was240/221, or about 1.09 times the uplinkfrequency. Kayton, Navigation, 320.

29. Kayton, Navigation, 320.

30. Duncan, “Apollo Navigation, Guidance,and Control,” Proceedings, figures 23, 24,and 25.

31. Paul Ceruzzi, A History of ModernComputing, second edition (Cambridge,Massachusetts: MIT Press, 2003),122–124.

32. It was a task that IBM itself failed atleast once in doing; see Frederick P. BrooksJr., The Mythical Man-Month: Essays onSoftware Engineering (Reading,Massachusetts: Addison Wesley, 1975).

33. Apollo 8 Flight Journal, transcript,http://history.nasa.gov/ap08fj/index.htm,accessed 25 January 2010; 54–55.

34. The command was “P-52”: a two-digit“verb” that the AGC interpreted to mean torealign the platform based on the star sight-ings.

35. Apollo 8, transcript, 67, MissionElapsed Time (MET) 000:56.

36. Apollo 8 Flight Journal, http://history.nasa.gov/ap08fj/index.htm accessed 25January 2010, also Robert Godwin, editor,Apollo 8: The NASA Mission Reports(Ontario, California: Apogee Books, 2000),178–179.

37. Apollo 8, transcript, 158, MET 5:08 andfollowing.

38. Apollo 8, transcript, 158, MET 5:36.

39. Apollo 8, transcript, 273, MET 17:53:25.

40. Apollo 8, transcript, 1056, MET123:31:14.

41. Kraft, Flight, 298. According the Kraft,the decision was his, after some dissentamong other NASA controllers.

42. Kraft, Flight, 200; MET 9:54:35.

43. Goodman, “Apollo 13 Guidance.”

44. Note that this did not in any way dimin-ish the importance of the AGC for controlfunctions: that is, putting the craft into theproper attitude, or guidance functions: thatis, ensuring during a rocket burn that thethrust vector is properly aligned with thecraft’s center of gravity as desired for agiven trajectory.

Q U E S T 17:4 201018

One of the NASA 26-meter MSFN antennas collocated with similar 26-meter, and muchlarger 70-meter antennas of the Deep Space Network. The DSN was developed foruncrewed missions far beyond the orbit of the Moon. A link established between the DSNantennas and the Manned Space Flight network, and this link was pressed into serviceto obtain live television images of the Apollo 11 landing, as well as emergency communi-cations during the Apollo XIII missions. Credit: NASA