Twentieth-Century Longitude: When Greenwich Moved

Introduction

A visitor to the Royal Observatory at Greenwich today will see a meridian line clearly marked out on the cobblestones of the courtyard. This “prime meridian” is aligned with an astronomical telescope, known as a transit circle, installed by the Astronomer Royal, George Airy, in the mid-nineteenth century. By the end of that century, the Greenwich meridian (illustrated in Figure 1) had come to define place and time globally. For place – meeting the needs of navigation and cartography – it was the most widely used reference for zero longitude. For time, it was observations made with Airy’s transit circle that defined Greenwich Mean Time, the reference for the world’s developing system of time zones. ¹

OPEN IN VIEWER

Figure 1.

The Airy transit circle, ready for use, behind the meridian line as currently marked in front of the Royal Greenwich Observatory.

Since then, however, practitioners seeking to determine accurate position or time have abandoned the idea of a geographically fixed prime meridian. That same visitor, if he were carrying a modern Global Positioning System (“GPS”) receiver, would note that it shows zero longitude to be about 100 metres east of the nineteenth century meridian. And if he were to return some years later, he would note that the GPS meridian had again moved a few centimetres. Such curiosities are the consequence of a rupture in our conventions for longitude in the 1980s, which followed the use of new techniques for distance measurement. But the origins of the rupture can be traced to much earlier in the twentieth century, when techniques of wireless transmission were first applied to the astronomic measurement of longitude. Wireless techniques are an important chapter in the history of longitude, but have received very little historical attention. ² I describe them, and the consequences of their use by astronomers, in this article.

To determine the relative longitude of two locations astronomically, it is necessary to determine their local time by observation of star transits, to preserve that local time with observatory clocks, and then to compare those local times to measure the difference. Wireless, an innovation of the turn of the twentieth century, allowed that comparison of times by the simultaneous reception of a single radio signal. In centuries past there had been many other techniques for the establishment of simultaneity: the immediate predecessor of wireless was cable telegraphy, which had been employed over the second half of the nineteenth century to establish a global longitude network constructed from hundreds of individual determinations. This activity all supported the increasing need for precision in longitude measurement. Initially, the demand had come from the practical needs of navigation and cartography. In due course, astronomical measurements of longitude were brought together with geodetic ones – all very much at the frontier of precision measurement – to develop knowledge of the size and shape of the earth (known as the “figure of the earth”). This was one of the grand projets of eighteenth and nineteenth century science, considered vital to the furtherance of the astronomical and mathematical sciences. ³

Greenwich was firmly fixed as the meridian of origin for all these measurements; its astronomical and navigational importance was practically reinforced by British dominance in submarine cables, allowing its connection for longitude measurements to the Continent, the Americas, and across the Empire. This network of cables allowed the creation of what has been called an “Electric Worldmap” with Greenwich at its centre. ⁴ Wireless telegraphy, nonetheless, promised improvement. In its essentials, wireless did exactly the same thing as cable, in allowing the establishment of simultaneity. At sea, of course, the absence of the need for physical connection was transformational: the reception of time signals by ship-borne wireless brought much greater certainty to determinations of longitude, until then reliant on the accuracy of chronometers. But on land, too, wireless came to dominate the measurement of astronomic longitude.

The twentieth century had brought yet further demands for precision in longitude measurement. Astronomers needed precision to ensure coherence of data obtained from observatories at different longitudes, thus improving measurement of time and allowing investigation of irregularities in the earth’s motion. Physicists required these better standards of time in many fields, for example, to establish the values of fundamental constants such as the speed of light. And geodesists had proposed new theories of continental drift in the early years of the century, which could only be tested by measurement of very small changes in longitude between continents. Wireless telegraphy was brought to bear on these problems, both as part of the development of a universal standard for time, which started before the First World War, and in three Global Longitude Operations of 1926, 1933, and 1957. Even though wireless techniques were not, as I will explain, intrinsically better than cable telegraphy, precision improved. Errors in longitude determinations were reduced from the order of a 10th of a second at the beginning of the twentieth century, to thousandths of a second during the era of wireless. ⁵

The historiography of precision measurement makes it clear, however, that precision relies on a great deal more than an individual using a carefully constructed instrument; rather, it requires an “extensive set of agreements about methods, materials, methods and values …” That is perhaps particularly true of observatories when considered as sites of knowledge production. ⁶ As Simon Schaffer has argued, the astronomer was in no way “communing directly with an unmediated heaven.” Even in Victorian times, observatories had become “sub-specialised workshops, stocked with batteries, induction coils, magnetometers, telegraph wires, spectroscopes, reagent bottles, Bunsen burners, photographic studios …,” and the twentieth century would bring much more matériel. ⁷ In seeking to explain improvements in the measurement of twentieth century longitude we need, I will argue, to look beyond the single prominent, yet ethereal, innovation of wireless. I will show that just as important were less obvious, but multifaceted, improvements in the complex techniques and apparatus of observation, recording, and computation employed within the observatory.

I will also show how wireless, and its use for time signals, caused us to re-think our conventions of longitude. Historically, Airy’s transit instrument at the Greenwich Observatory defined Greenwich Mean Time and the location of the prime meridian. From the late 1920s, wireless techniques allowed the development of a new global standard of time, calibrated by averaging time determinations from several contributing observatories, each adjusted for its conventional difference in longitude from Greenwich. The inevitable errors in such a process meant that this time was always fractionally different from time actually observed at Greenwich, and by an amount that varied from year to year. In other words, the world’s transit instrument of reference had moved from Greenwich to a fictional mean observatory at a longitude slightly adrift from Greenwich. Longitude was, as one practitioner put it, the “dustbin” into which all the various errors were put in order to “stick together” our measures of time. ⁸

By the 1950s, there was the further complication that Greenwich actually had moved, in that the Observatory had been relocated well away from the pollution of London. Measurements were therefore no longer possible at the meridian itself. As a consequence, in the early 1960s, a theoretical mean observatory, very slightly distant from Greenwich proper, formally superseded Greenwich as the conventional zero of longitude. And, in the following years, the web of astronomic time and longitude measurements which embraced that theoretical meridian underwent constant refinement and adjustment in the search for improved precision. Greenwich, and much else besides, continued moving without any clear destination. Then, in the early 1980s, new techniques of distance measurement, using lasers and satellites, replaced traditional astronomical longitude determination and the world’s longitude network was redefined. With this rupture, Greenwich moved again.

I will conclude by interpreting the mobility of zero longitude within the historiography of scientific progress. The story of measuring longitude is that of a search for ever-increasing precision, which we understand to mean the reliability or repeatability of particular measurements. The improvement took place, I will show, through the continual refinement of a global web of measurements of considerable complexity. But as Norton Wise has argued, “Problems of establishing precision therefore become simultaneously questions of establishing agreement within a community. Precision requires standardisation.” ⁹ And the standard for zero longitude was, by the beginning of the twentieth century, the meridian through Airy’s transit circle at Greenwich. Standards, however, are not immutable. Hasok Chang has argued that improvements in standards are driven by an imperative of progress, but constrained by a principle of respect between the prior and later standard. ¹⁰ For example, our historic standards of length based on metal bars were superseded during the twentieth century by ever more precise standards based on the wavelength, and then the velocity, of light. These new standards clearly respected the old because the changes were effectively undetectable to practitioners. ¹¹ Such continuity was not at all the case for twentieth century standards of longitude. But I will explain how, despite the mobility of our reference meridian, we do still respect that original line in the courtyard at Greenwich.

The beginnings of long-distance wireless

Wireless transmission was an innovation of the last years of the nineteenth century. France was among the leaders in its development, for reasons both financial and military. The then dominant technology for long-distance communication was cable telegraphy; that was in turn dominated by the British, who controlled over three-quarters of the world’s 300,000 kilometres of cable. France, in contrast, had to make do with little more than a couple of connections across the Atlantic and the Mediterranean. ¹² The financial consequence was that it cost millions of francs a year for France to communicate with those large parts of its Empire in Africa, Asia, and the West Indies to which there was no direct French telegraphic connection – millions begrudgingly paid to British cable operators. The military consequence was that the British could intercept, censor, or even deny access to foreign users of their network at will. For example, during a period of Anglo-French sabre-rattling in the Sudan in the 1890s, the British troops remained in continuous contact with London, while the cable connection used by the French troops mysteriously went silent. Yet worse, Britain owned most of the specialist cable-laying ships that could grapple and cut cables at sea. France therefore felt insecure in its global communications, and the new technology of wireless offered an immediate solution. ¹³

Military interest in wireless telegraphy was apparent from the very start. The pioneer Guglielmo Marconi arranged a demonstration of cross-channel transmission in 1899; the French observer was Captain Gustave-Auguste Ferrié, a military engineer who was to play a central role in the application of wireless to measurement of time and longitude in the years to come. ¹⁴ He was both technically and practically skilled, making a name for himself by leading an expedition in 1902 to establish wireless communication with the island of Martinique after a volcanic explosion severed its cable connections. And it was he who identified the Eiffel Tower, built for the 1889 World Fair but redundant by the early years of the twentieth century, as an ideal support for the long antennae then essential for long-distance transmissions. Wireless equipment was therefore installed at the Eiffel Tower as early as 1903, illustrated in Figure 2.

OPEN IN VIEWER

Figure 2.

The first Eiffel Tower radio installation. This basic configuration, with four antennae stayed by trees, was erected in 1903. It was superseded a few years later by a larger array, and the radio equipment was moved from the small hut (marked P) to more substantial premises underground. One reason for this move was that an inquisitive passer-by could otherwise hear the signals being sent from the noisy spark transmission apparatus.

There was a contemporaneous French proposal that wireless could be used to transmit time signals to ships several times a day, in order to reset their chronometers and hence improve their determinations of longitude. The limit of radio range was at the time only some 500 kilometres, but it was confidently predicted that this was capable of improvement. ¹⁵ And while it was admitted in France that the issue of wireless time signals was essentially an international one, the Eiffel Tower transmitter was the obvious place to start. ¹⁶ The Bureau des longitudes therefore proposed in 1909 that a wireless time signal service be established as soon as possible, specifically for the purpose of longitude determination. The Ministre de la Guerre provided finance for new high-powered transmission equipment, from which six new long-wave antennae were attached to the Tower. And the Paris Observatory provided an electrical time signal for the transmitter. The resulting transmissions, which commenced in 1910 with just one daily signal at a fixed time, could be received over a large part of Europe and the North Atlantic. Initially, these time signals, known as signaux ordinaires, were used mainly by shipping. The precision of reception was limited to perhaps one quarter of a second, adequate for navigational purposes. A couple of years later, additional signaux scientifiques were added, intended to be used for more demanding astronomic and geodetic purposes, and said to be detectable with a precision of one-hundredth of a second. ¹⁷ The Eiffel Tower transmitter was intended to be, as one historian has described it, “the greatest time synchronizer in the world.” ¹⁸

Other countries with an imperial or naval need were building long-distance wireless transmitters in the years before the First World War, and a couple – the United States and Germany – started time transmissions. ¹⁹ But the time signals from the Eiffel Tower were not just among the first; they were almost certainly the most accurate, as a result of the techniques used at the Paris Observatory. For the determination of time, the renowned Great Meridian Circle was soon dedicated exclusively to the purpose, with an extensive programme of daily astronomical observations and regular verification of the various instrumental constants. For the preservation of time, the several observatory clocks included some of the most advanced available, operating at controlled pressure and temperature to minimise error. In fact, the greatest risk to accuracy was bad weather, which could prevent astronomical observations for some days and therefore allow a slight drift of the transmitted time. Accordingly, it was arranged that other French observatories would compare the time of the Paris signal against their own time observations and communicate the results. Greenwich joined in too, sending its observations to Paris by daily postcard. ²⁰ These time transmissions from the Eiffel Tower were soon at the centre of two closely inter-related projects which I will describe. The first was the application of wireless techniques directly to the establishment of longitude. The second was the attempt to establish a single measure of time globally at the very highest level of precision.

Longitude by wireless before the First World War

For the measurement of longitude, the innovation of wireless was greeted enthusiastically by its proponents, who extolled its simplicity, speed, and precision. ²¹ A first test measurement was carried out in 1910, using the Eiffel Tower transmitter, between the Paris Observatory and the nearby observatory of the Service géographique de l’Armée at Montsouris. ²² A longitude measurement between Paris and Brest followed, a distance of about 600 kilometres; the result was said to be highly satisfactory. ²³ The next step was a measurement between Paris and Bizert in Tunisia, at a distance of over 1500 kilometres. Here, there were further refinements in technique. A transmitter was established at Bizert as well as Paris allowing a two-way measurement, which reduced uncertainty as to the transmission time of the wireless signals. In a continuation of the best telegraphic longitude techniques, independent measurements by two pairs of observers were made simultaneously. And for time determination, a new technique for automatic recording of stellar observations, a device known as the self-registering micrometer, was employed. Previously, an observer noted the moment that a star crossed the vertical meridian thread in his telescope and manually made an electrical contact to record the event. Now, with an electrically driven micrometer, he could “follow” the star with the sighting thread, while contacts were made automatically. Errors from what is known as the personal equation (the tendency for individual reaction times to differ) were much reduced. ²⁴ Precision of the order of one-hundredth of a second in a longitude determination was now reported.

Another perspective on the benefits of wireless came from one of France’s leading military geodesists:

The use of wireless for the determination of differences of longitude is of vital importance for colonial geography; its use allows the establishment, with much greater precision than by transport of chronometers or astronomical observation, of the position of fundamental points in those parts of the world, still vast, without telegraphic connection. ²⁵

Mapping of new colonial territories could now be done by the determination of a number of fundamental geographical points, spaced at distances of around 50 kilometres from each other and then joined by triangulation. It was said that the only apparatus needed for this was a portable wireless receiver, a collapsible antenna, a sighting instrument for determining local time and latitude, and a chronometer. And all this was apparently easily transportable by mule or even on the back of a man. The King of Belgium, whose revenues from the Congo would have given him a keen interest in colonial geography, was given a demonstration in 1911: the longitude of the Royal Palace in Brussels was measured in what were described as “colonial conditions,” using the apparatus and techniques that would be employed in the field. ²⁶ French military surveyors were using wireless in Africa very soon afterwards, in the colonies of Senegal and Mauretania, and to re-arrange some borders between French and German colonies. ²⁷

Next, the increasing power of wireless transmitters re-opened the question of transatlantic longitudes. The United States Naval Observatory in Washington was to provide time signals for a new high-power transmitter nearby at Arlington, Virginia. Because those signals were to be used to determine the longitude of other observatories in the United States, and because its transmissions could overlap with those from Paris in mid-Atlantic, accurate knowledge of its longitude was essential. ²⁸ The longitude of Washington had, however, only ever been deduced indirectly, and somewhat unsatisfactorily, from older transatlantic cable measurements terminating at another American observatory. A proposal to re-measure Paris–Washington longitude therefore came from the Director of the Naval Observatory in 1912, and work swiftly commenced. Given that the distance between Paris and Washington was some four times greater than any previous longitude determination by wireless, there was no certainty that it could be done. A French team visited Arlington in early 1913 to carry out preliminary experiments and to familiarise the American observers with techniques that were new to them; they found the transatlantic signals from Paris to be faint but sufficiently clear for the purpose. The definitive measurement was started later the same year and took place over the winter months (which are better for wireless transmission). And the exercise was considered a resounding success, with uncertainties in the outcome measured in thousandths of a second. One American writer suggested that “the scientific world will look upon this longitude campaign as not only an epoch-making contribution to this field of science, but as a model for future achievements.” ²⁹

Improvement, however, wasn’t much to do with wireless. The basic techniques employed would have been completely familiar to a participant in any of the better telegraphic longitude determinations of the previous 40 years. ³⁰ And the use of wireless transmission was in one respect a step backwards, in that it brought greater uncertainty as to the electrical characteristics and the length of the signal path, compared to that of an undersea cable. So where had the improvement come from? What is emphasised in contemporary accounts is the contribution of a number of fundamental improvements in observatory technique. ³¹ The first was the use of the self-registering micrometer, as seen in the Paris–Bizert determination. Second was the installation of what was known as the Riefler clock in Paris and Washington, which ran at a more stable rate than any observatory clock before. The device used a pendulum made from invar, a new alloy of very low temperature expansion coefficient, kept in a partial vacuum. ³² Third was the employment of new American star catalogues, allowing the observation of a greater number of stars than had been customary at Paris, their positions carefully selected to minimise instrumental errors. ³³ And fourth was the use of two different techniques for the measurement of the time of radio signals: one by ear, using the method of coincidences and another with a new chronograph using photographic techniques to create a permanent record of signals and clock beats. Longitude determination benefitted, therefore, from improvement in observatory techniques on a very broad front.

The very few wireless longitude determinations made in the early years of the twentieth century, mainly by France and the United States, were trivial in their scope compared to the global network of longitudes established by telegraph in previous decades. ³⁴ Ferrié therefore conceived a much more ambitious project. This would start with the establishment of a first-order network in the Northern hemisphere, requiring the measurement of the longitude of nine stations around the world all situated close to the 45th parallel; this network would act as reference for the deduction of the longitude of further locations, and the exercise could all then be repeated in the southern hemisphere. ³⁵ The Bureau des longitudes approved the project and obtained financing. Operations started in mid-1914 with a measurement of the first link of the chain, between Paris and the observatory of Poulkova. The outbreak of war then brought the project, and all other wireless longitude measurements, to an end for over a decade.

Standards of time before the First World War

If the transmission of time signals by wireless began to transform the measurement of longitude before the First World War, the question remained as to exactly what time was being transmitted. During the later nineteenth century, there had been much progress globally in the creation of standard times. This, however, was a story of standardisation of time for general rather than precision use, driven in large part by the spread of the railway and the telegraph, and has been well told elsewhere. ³⁶ By the end of the century, most countries had a single national time, or a series of time zones, deduced from Greenwich Mean Time. France had gone so far as to adopt Paris time for national use, but remained one of the few to hold out against relating national time to the Greenwich standard.

The reason for this intransigence was in large part a matter of amour-propre, because any linkage to Greenwich time opened up the wider issue of acknowledgement of the Greenwich meridian as the global geographic standard. That risked the demise of the Paris meridian, the basis of French cartography and naval charting for centuries. So when the Eiffel Tower first started transmitting time signals in 1910, they were of French national time as determined at the Paris Observatory. A bill to align French time with Greenwich time had been stuck in committee for years, but the success of the Eiffel Tower transmitter seems to have re-invigorated the legislative process. ³⁷ So in early 1911, it was decreed that legal time in France was to be Paris mean time retarded by 9 minutes 21 seconds – that being almost exactly the difference in longitude between Paris and Greenwich. ³⁸ French legal time was, of course, Greenwich Mean Time in all but name. So the Eiffel Tower was soon transmitting its version of Greenwich time to a large part of the world. Any difference between what one might call proper Greenwich Mean Time and the Parisian simulacrum (in fact, less than one-tenth of a second) would have been concealed by measurement uncertainties. Nonetheless, this was conceptually an important step: for an extensive international audience, time was being defined by a transit instrument in Paris, as corrected by a conventional longitude difference, and not in Greenwich. This was, in an operationalist sense, the introduction of a new kind of time. ³⁹ Greenwich time was no longer coming from Greenwich.

France now moved to internationalise the standardisation of time, and the Bureau des longitudes called a Conference internationale de l’heure, held in Paris in 1912. The stated purpose of this conference was to study the means for the practical realisation of the unification of time (although one more sceptical delegate noted that “a weighty reason” for the French calling this conference was that they wished to maintain the pre-eminence of the Eiffel Tower signals). ⁴⁰ There was unsurprisingly a French president, Guillaume Bigourdan of the Paris Observatory, supported by Ferrié as general secretary. French ambition aside, the collective objective was that identical, and ever more accurate, wireless time transmissions should be receivable across the globe. Such coherence could not be achieved by a series of national time transmitters taking time from their own observatory, each inevitably subject to local errors. Accordingly, the French proposed the creation of an international time service which would bring together time observations from participating observatories and produce a single universal time. This standard time would then be distributed for transmission by all. ⁴¹

Practitioners were clear, however, that time and longitude were inextricably linked. A second conference was held in Paris in 1913, to seek intergovernmental approval for execution of the project and the establishment of a Bureau international de l’heure (“BIH”). At this conference, a letter from the Association géodesique internationale, the body that for the last 60 years had co-ordinated longitude measurements, was read to all the delegates. Its President expressed wholehearted support for the proposed system of unified time signals. He went on to write that “what interests us is that they could also give data about the differences in longitude between observatories,” values which, he explained, were unfortunately still corrupted by the many types of error that he enumerated. ⁴² War, however, intervened and the convention that concluded the 1913 conference was never ratified. The Bureau international de l’heure was thus denied formal international status, although it came into being as an organ of the Paris Observatory, which continued to supply time signals to the Eiffel Tower transmitter. ⁴³ And the ambitious proposals for the establishment of a worldwide network of wireless stations for the transmission of time signals (illustrated in Figure 3) came to nothing. But it was this global approach to the measurement and distribution of time that would next be applied to the determination of longitude by wireless.

OPEN IN VIEWER

Figure 3.

The Time Signal Wireless Telegraph stations proposed, but mostly not realised, at the Conférence Internationale de l’heure of 1912 (edited extract from a global map, excluding the Americas). The British stations were to be Cape Town, Colombo, and Hong Kong.

Longitude by wireless after the First World War

The utility of wireless was proved beyond doubt during the war; it was perhaps appropriate that news of the Armistice was transmitted from the Eiffel Tower, which during the conflict had carried messages containing a total of over 10 million words. ⁴⁴ The war had also brought much technical progress, although effective long-distance use of wireless still remained very much a matter of brute force: big aerials and powerful transmitters were essential. France remained in the vanguard. A new transmitter had been installed during the war at Lyon, reported to be as powerful as any in the world, as a reserve for the Eiffel Tower and with double its range. It was followed by an even bigger transmitter at Bordeaux (illustrated in Figure 4); this was built with a great deal of help from the United States, following their entry into the war. It comprised eight 250-metre high towers, and was 10 times again more powerful than that at Lyon, communicating across the Atlantic in conjunction with another huge transmitter at Annapolis, Maryland. ⁴⁵ Ferrié had been in charge of all French military radiotelegraphy throughout the war so was well placed thereafter to resuscitate his project for the measurement of global longitudes – the execution of which had been simplified by the increasing range of transmitters, which could now reach half way round the world. Transmissions from France, for example, could now be received as far away as Shanghai and New Zealand. ⁴⁶

OPEN IN VIEWER

Figure 4.

The Bordeaux transmitter building and six of its eight aerial towers, which when built made up the most powerful wireless station in the world. It was named Lafayette, after the French Marquis who fought alongside America in the revolutionary war. Construction having been started in 1918, it was inaugurated in 1920 and destroyed during the Second World War.

The Bureau des longitudes proposed a global longitude project in 1919, but international agreement was slow in coming. ⁴⁷ That was in part because of the complex architecture of post-war science: two newly established bodies, the International Astronomical Union and the International Union of Geodesy and Geophysics were both interested in the project, and established a joint commission chaired by Ferrié. ⁴⁸ There followed lengthy technical debates, and it took some years to obtain final approval. The project was finally executed in late 1926, with the longitude network to be measured with “all the precision made possible by modern astronomical and wireless techniques.” ⁴⁹ The first objective was the elimination of the inconsistencies known to be remaining in the global network of longitude. The second was to lay the foundation for a test of new theories of what was known as continental drift, which postulated that positions on the surface of the earth were not stable; for this, a re-measurement was planned after an interval of a few years. ⁵⁰ Ferrié saw clear opportunity for improvement, because the execution of this World Longitude Operation would allow something that had been impossible with telegraphic techniques. The world’s telegraphic longitude network had been patched together over decades, from hundreds of individual measurements using all manner of apparatus and observational practices. In contrast, the World Longitude Operation would establish a coherent network with one set of simultaneous astronomic and wireless measurements, the techniques and computation of which could all be carefully managed. It attracted public attention, as shown in Figure 5.

OPEN IN VIEWER

Figure 5.

The first World Longitude Operation, as described in the New York Times, 6 September 1925. (Cited by Oreskes, The Rejection of Continental Drift (Ref. 53)).

For this first World Longitude Operation, three fundamental stations were chosen, at roughly the same latitude and spaced about 8 hours apart: the French observatories at Algiers and Shanghai, and the United States Naval Observatory at San Diego (though the French wanted observers at the latter). ⁵¹ The details of the project evolved over time, but in its final form, there were about 20 further first-order stations (including the observatories at Paris, Greenwich, and Washington). These would form an around-the-world polygon of observatories whose longitude was determined to the highest level of precision. A number of widely dispersed secondary stations brought the total number to about 40. The stations were initially planned to receive signals from just three time signal transmitters: Annapolis and Bordeaux (as constructed for the wartime transatlantic link) together with Honolulu (Pearl Harbour). These were powerful enough to cover the globe, even though other transmitters were added. ⁵²

Thus was built a new global longitude network over about 2 months of observations in late 1926, using techniques that built on the pre-war practice, with some improvements. Perhaps the most important innovation was the use of an instrument known as the prism astrolabe which offered a different and potentially more precise method of astronomical time determination than the transit instruments employed to date. ⁵³ At the fundamental stations, the astrolabe was used as well as the traditional transit circles. In addition, the operation employed the latest Eichelberger star catalogue, named after and created by the director of the United States Nautical Almanac Office. And the entire measurement process was operated under strict protocols designed to ensure consistency of results. The apparatus at the secondary stations was inevitably somewhat less sophisticated than at the first-order ones, as shown in Figure 6, but was nonetheless of high quality.

OPEN IN VIEWER

Figure 6.

A secondary station in the first Global Longitude Operation at Manila, the Philippines (then under American control). The transit instrument mounted on a stone pillar seen at the left is of a type known as the Bamberg broken-transit telescope, an advance on the traditional reversible transit instrument, in common use with the US Coast Survey (see Swick, World Longitudes (Ref. 54)). The wireless receiving equipment is behind and the weight-driven chronograph for recording signals is to the right.

The first World Longitude Operation was followed by a second, employing some minor improvements in technique, in 1933. ⁵⁴ We now know, however, that while the probable errors of longitude determination in these operations were of the order of metres, the rate of continental movement is of the order of only centimetres per annum. So, given the brief lapse of time in geological terms, the second World Longitude Operation produced no conclusive evidence of the existence of continental drift. ⁵⁵ As one American geodesist said, “The results really didn’t show anything.” ⁵⁶ The measurement operations were, rather, a demonstration of a clear shift in technique. The days had passed when bilateral longitude measurements, using cable or wireless signals to create simultaneity, were of any use at the frontier of precision. That precision now came from multiple simultaneous measurements, brought together statistically in order to manage error and to achieve internal coherence, creating a global network of astronomically determined longitude. There was, however, one aspect of historic longitude determinations that did remain unchanged: all the results were still referred to a meridian of zero longitude at Greenwich. It was this particular certainty that wireless techniques were soon to undermine.

The end of astronomic longitude

The mobility of this mean meridian was at this stage more of a curiosity than any real difficulty for practitioners. Geodesists didn’t readily change their points of reference: an Ordnance Survey map, for example, still showed the prime meridian going through the observatory at Greenwich, just as it had done since the first triangulation of Great Britain, made a century before. ⁶⁶ The Second World Longitude Operation had employed l’heure définitive as time standard, and thus a mean meridian within its calculations, but its final results were still referred to Greenwich proper. ⁶⁷ And even when wireless techniques showed up material inaccuracy in the conventional value of a particular observatory’s longitude, revision was rare: the difficulties of unravelling decades of historic data and the need to minimise discontinuities in the position of the mean observatory (and thus in the measure of time) militated against change. ⁶⁸ There followed, however, a series of further improvements in technique, the impact of which was clearly seen in a third Global Longitude Operation.

The period from 1 July 1957 to 31 December 1958 was known as International Geophysical Year, which coincided with the first satellite launches. This vast project was contemporaneously presented as a peaceful collaborative scientific endeavour, but was in fact closely connected to the Cold War military and national security objectives of the leading participants. It addressed a range of disciplines in the sciences now known as geophysics – meteorology, geomagnetism, seismology, gravimetry, oceanography, and much more. ⁶⁹ It included the third Global Longitude Operation, similar in its essentials but more ambitious than those that went before. This time it lasted over 2 years, rather than a couple of months. There were over 50 stations at which 1 million time signals and 40,000 astronomical observations were recorded – some 20 times more than in the first operation. And there was technical improvement on a broad front. ⁷⁰ For time determination by astronomical observation, a new device known as the Photographic Zenith Tube (“PZT”) brought a significant enhancement in precision. ⁷¹ For the preservation of time, mechanical clocks gave way to quartz clocks and atomic clocks. And radio techniques had progressed too, with short wave rather than long-wave transmission now being preferred. The published results showed the longitude of each station to the nearest one-thousandth of a second, compared in every possible case to the determinations from the first and second longitude operations in order to test again the theory of continental drift. This time there was a conclusion: there was a “probability” that Europe and North America were getting closer. ⁷² But the comparisons of longitude upon which this conclusion were based were far from straightforward, requiring wholesale reworking of the older observational data. Why was this necessary?

The third Global Longitude Operation had to cope with two major developments since the earlier ones. The first was that Greenwich actually had moved. At Greenwich, light and air pollution from London was making it difficult for the Royal Observatory to operate effectively, so in the 1950s, it was moved to the countryside, at Herstmonceux in Sussex. The location of Herstmonceux relative to Greenwich had been determined somewhat imperfectly, and by the time of the third Global Longitude Operation, there was no suitable observing apparatus at Greenwich itself. ⁷³ So there was little alternative but to work with a fictional mean observatory. ⁷⁴ The second was that standards of time had changed. It had long been known that there were minor irregularities in the earth’s rotation, including a phenomenon known as polar motion, which involves a movement of the earth’s axis of rotation with respect to its surface. Although only of the order of metres, the resulting variations in longitude of points on the earth’s surface distort measurements of time. Traditionally determined astronomical time was now corrected for polar motion: in the new terminology of the 1950s, UT0 (UT for Universal Time) was supplemented by a corrected variant called UT1. ⁷⁵ An additional discontinuity between the first two and the third longitude operations came from the use of a new and improved star catalogue. It followed that, for comparability, longitudes from the first two operations had to be corrected retrospectively for both polar motion and the new star catalogue. ⁷⁶ All this further undermined certainty as to historic longitude determinations.

Such instability followed through into the work of the BIH. As I have explained, it had been practice for the conventional longitude of contributing observatories to remain unchanged, even though many had been ascertained decades before by methods long since superseded. By the early 1960s, it was clear that some of these longitudes were significantly in error. Individual amendments were, however, formally discouraged by the International Astronomical Union on the basis that they would damage the integrity of the global network. ⁷⁷ The proposed use of a new star catalogue from the start of 1962 was, however, going to introduce a discontinuity in any event. The BIH took the opportunity for a wholesale revision of the longitudes of the 40 contributing observatories. The new values incorporated data from the recent Global Longitude Operation, and were now formally referred to a zero of longitude at a mean observatory approximating Greenwich, rather than Greenwich itself. The change in reference meridian resulted in a small discontinuity in time measurement, less than one-hundredth of a second; an undesirable, but necessary, consequence of improving the network as a whole. ⁷⁸ Further modifications to the conventional longitudes of the by now 70 contributing observatories came in 1968; this time, they resulted from changes to the convention for correcting for polar motion, and from amendment of the value of a particular astronomical constant. ⁷⁹ This search for improved precision meant that the relative longitudes of Greenwich, its actual replacement at Herstmonceux, its fictional replacement by a mean observatory, Paris, and pretty much everywhere else, continued changing during the 1960s. ⁸⁰

This process of evolutionary improvement took place against a background of important developments in our standards of time. Until the 1950s, a second was simply a natural standard: 1/86,400th of a day, as measured by astronomers and based on the assumption that the earth’s rotation was uniform. In fact, it had been suspected since the nineteenth century that it is was slowing, and that was proved during the 1930s. ⁸¹ As the frontier of precision in astronomy and the physical sciences moved forward, and as the new Système Internationale of units was developed, this “astronomical second” was no longer adequate. A better definition of the second “satisfying the most rigorous needs of science” was needed. ⁸² Astronomers, reluctant to cede primacy to physicists, created a second based on what they called Ephemeris time, based on the orbital motion of the earth around the sun. The Ephemeris second, cumbersome to realise, was soon superseded in 1967 by a definition (of notionally identical length to the Ephemeris second) based on transitions of the caesium atom and realised with atomic clocks. These were in operationalist terms, different kinds of time, based, respectively, on measurement of the rotation of the earth, the motion of the earth round the sun, and the period of an atomic transition. These different kinds of time, and definitions of the second, gave rise to a variety of different timescales. Among them, practical timekeeping is currently based on an atomic timescale known as TAI (Temps Atomique Internationale). For astronomers, though, the need for accurate knowledge of the Earth’s rotation and orientation in space means that UT1 endures. There is however, conflict here: the atomic second and the astronomical second differ slightly, so TAI and UT1 diverge. To accommodate this difficulty, a third standard known as UTC acts as our practical civil standard of time, approximating UT1 to within no more than nine-tenths of a second by the insertion of “leap seconds” into TAI. ⁸³

The link between time standards and the rotation of the earth was thus partially broken, but that had little effect on determinations of longitude. Until 1983, the BIH continued regularly to publish longitudes for contributing observatories, using conventional astronomic techniques. Change came instead from innovations in distance measurement, such as laser ranging to satellites and the moon, VLBI (Very Long Baseline Interferometry), and the US Transit Doppler satellite navigation system. Initially used to supplement the older techniques, they in due course replaced them: in 1984, the BIH published an entirely new geodetic terrestrial reference system based on a network of about 30 sites. In operationalist terms, this was a fundamental change, a new kind of longitude, to be defined by distance and not by time. Underlying this, our standard of length is defined by ascribing a conventional value to the velocity of electromagnetic radiation and by the standard atomic second. So we actually determine distance by measuring time differences in terms of that second. ⁸⁴

Geodetic measurement of longitude is of course nothing new and predates electromagnetic distance measurement techniques by many years. During the eighteenth century, the principal technique for determining the figure of the earth was the measurement by triangulation survey of the length of meridian arcs at various latitudes, together with the astronomic determination of the latitude subtended. But during the nineteenth century, and especially after the employment of the telegraph to assist measurement of astronomic longitude, similar techniques were applied to the measurement of long arcs of parallel. ⁸⁵ Geodetic and astronomic longitude are, however, not identical because of a phenomenon known as deflection of the vertical. Geodetic longitude coordinates are based on a conventional spheroid, the dimensions and ellipticity of which are chosen to be a close approximation to the actual figure of the earth; the vertical at any location on that spheroid passes through its axis of rotation. Astronomic longitude coordinates are based on observations from a transit instrument whose verticality is established using apparatus such as a basin of mercury. The earth is, however, irregular in its topography and internal composition so it is usually the case that the apparent gravitational vertical does not pass exactly through the earth’s axis. This influences local time determinations made with the transit instrument. The deflection of the vertical at any location, in an East-West direction, is therefore the difference between astronomic and geodetic longitude. ⁸⁶

For geodetic definition of location to be effective, we need to define not only the dimensions of a conventional spheroid but also a meridian of reference. ⁸⁷ The meridian chosen by the BIH, now called the meridian of the International Terrestrial Reference Frame (“ITRF”), is the one that finds itself about 100 metres east of Airy’s transit circle. Yet we can still see the ghost of Airy behind this meridian’s location. There have been all sorts of conjectured explanations for the meridian’s shift, but recent research has proved that it was a simple, if accidental, consequence of the long-standing policy of the BIH to maintain continuity of time. It is the case at Greenwich, as at most other locations on Earth, there is a small local deflection of the gravitational vertical; in other words, its astronomical longitude and geodetic longitude do not coincide. So if we could press Airy’s transit circle back into service today, it would still show what one might call the “right” time: the gravitational deflection affecting the telescope at the Airy meridian and the longitude offset from that to the ITRF meridian cancel each other out. ⁸⁸ As the director of the BIH later wrote, the shift “could have been avoided by different means, but nobody considered the question in due time.” ⁸⁹ Once again, longitude was the dustbin into which errors were put in order to stick together our measures of time.

Conclusion

The purpose of this article has been to describe the twentieth century innovations that were applied to improve the measurement of longitude, together with the changes that resulted to our conventions of longitude. I would like to draw out, in conclusion, some connections between this history, and the historiography of the nature of change and improvement.

I have explained that the establishment of astronomic longitude requires the determination of time, the conservation of time, and the comparison of time. Changes in technique for the last of these, in this case wireless, were the most visible innovation of the early and mid-twentieth century. It was techniques for the first two, however, which were the foundation of precision. Change here was often more gradual, discrete, and multifaceted. It encompassed improvement in areas such as instrumentation for astronomical observation, management of the personal equation, techniques of data recording, methods for the combination of observations, observatory horology, and so on. These have been well described as “observatory techniques,” the whole set of physical, methodological, and social techniques rooted in the observatory. ⁹⁰ It was further developments in observatory techniques, already refined by half a century of telegraphic longitude work, which contributed as much to improvement in precision as the application of wireless. I conclude that – as Donald Mackenzie has argued – proper analysis of exactly how improvement occurs must look beyond single visible innovations to changes in the often long-lasting, less prominent, supporting techniques. ⁹¹

Determinations of time and longitude were deduced from a worldwide web of measurement techniques of considerable complexity, the precision of which improved continuously during the wireless era. The longitude network created with the help of the electric telegraph in the latter part of the nineteenth century was initially strengthened as individual strands were re-measured by wireless. It was then improved by the first two Global Longitude Operations, refined by the work of the Bureau Internationale de l’Heure, tested by the third Global Longitude Operation, and modified further in the 1960s. It accommodated, sometimes with discomfort but never with rupture, changes of many sorts: examples include the periodic replacement of star catalogues; a 10-fold expansion in the number of contributing observatories and various corrections of their conventional values of longitude; incorporation of the effect of irregularities in the earth’s rotation, and even the abandonment of a fixed observatory of reference.

In terms of improvement, this is something more, I suggest, than what Thomas Kuhn called the “complex but time-consuming mopping-up operation” that constitutes the bulk of scientific practice. ⁹² It is much better described in the terms of Hasok Chang as an epistemic iteration, an ill-defined process of improvement in which later stages of knowledge are built upon earlier, correcting and refining them, but which cannot be deduced from them in any straightforward sense. ⁹³ An integral part of such iteration is an element of self-correction of measurement methods. From the late 1920s, we deduced our global standard of time, l’heure definitive, by combining local time determinations from many observatories, each adjusted for their conventional longitude from Greenwich. In turn, we corrected those conventional longitudes by comparing the local time observations with l’heure definitive, and so on. Such iterative improvement lacks the rigour of a mathematical iteration (which has a pre-determined final destination) but is nonetheless seen to be an effective method of what Chang calls coherentist progress. ⁹⁴

Finally, how should we interpret the mobility of zero longitude, one of the world’s most important measurement standards, in the twentieth century? As Chang has argued, there is an imperative for progress in the field of standards. ⁹⁵ Within this, investigations based on a prior standard can and do result in a new and improved standard. Indicators of improvement include the greater consistency of judgements reached by use of the standard, and the opportunity to achieve more than the old standard permitted. But that progress is constrained by a principal of respect: the new standard does not have unconditional authority over old. It should show sufficient agreement with what has gone before, but with enough liberality to give breathing space for progress. As one historian put it, there is a need to “to escape the past without doing violence to the historical relationship between the present and the past.” ⁹⁶ That is exactly why and how Greenwich moved. The mean observatory of the mid-twentieth century facilitated a better, more stable standard for time than one based on measurements at any single observatory: the slight mobility of that mean observatory just provided some necessary breathing space for the new standard’s development. The ITRS meridian of the later twentieth century, which serves as that for GPS, then supported new and improved techniques of longitude measurement, but again maintained respect for the past through the continuity of our standards of time. We can thus understand why Greenwich moved, and how, a century after it began to move, our standard does still respect that historic line in the observatory courtyard.