You’re reading Age of Invention, my newsletter on the causes of the British Industrial Revolution and the history of innovation. This edition went out to over 13,400 free subscribers. To support my work, you can upgrade your subscription here:
The other day I was tagged into answering the following question:
“Economic historians often write about why Ancient Rome didn’t industrialise with steam engines. I think a better question is why didn’t they invent the telegraph — something that to any emperor would have been of immediate, obvious importance”.
I am guilty as charged when it comes to writing about why it took so long for the invention of the steam engine. Possibly guiltier than most, given my recently-completed three-part series on the question (here are I, II, and III). But I’ve also long wondered the same about telegraphs — not the electric ones, but all the other long-distance signalling systems that used mechanical arms, waved flags, and flashed lights, which suddenly only began to really take off in the eighteenth century, and especially in the 1790s.
What makes the non-electric telegraph all the more interesting is that in its most basic forms it actually was used all over the world since ancient times. Yet the more sophisticated versions kept on being invented and then forgotten. It’s an interesting case because it shows just how many of the budding systems of the 1790s really were long behind their time — many had actually already been invented before.
[Public service announcement: before we get stuck in, this post is sufficiently long and illustrated that there’s a high risk that this email may end up truncated for some people. If this happens, you can always read the post online by simply clicking here.]
The oldest and most widely-used telegraph system for transmitting over very long distances was akin to Gondor’s lighting of the beacons, capable only of communicating a single, pre-agreed message (with flames often more visible at night, and smoke during the day). Such chains of beacons were known to the Mari kingdom of modern-day Syria in the eighteenth century BC, and to the Neo-Assyrian emperor Ashurbanipal in the seventh century BC. They feature in the Old Testament and the works of Herodotus, Aeschylus, and Thucydides,with archaeological finds hinting at even more. They remained popular well beyond the middle ages, for example being used in England in 1588 to warn of the arrival of the Spanish Armada. And they were seemingly invented independently all over the world. Throughout the sixteenth century, Spanish conquistadors again and again reported simple smoke signals being used by the peoples they invaded throughout the Americas.
But what we’re really interested in here are systems that could transmit more complex messages, some of which may have already been in use by as early as the fifth century BC. During the Peloponnesian War, a garrison at Plataea apparently managed to confuse the torch signals of the attacking Thebans by waving torches of their own — strongly suggesting that the Thebans were doing more than just sending single pre-agreed messages.
About a century later, Aeneas Tacticus also wrote of how ordinary fire signals could be augmented by using identical water clocks — essentially just pots with taps at the bottom — which would lose their water at the same rate and would have different messages assigned to different water levels. By waving torches to signal when to start and stop the water clocks (Ready? Yes. Now start… stop!), the communicator could choose from a variety of messages rather than being limited to one. A very similar system was reportedly used by the Carthaginians during their conquests of Sicily, to send messages all the way back to North Africa requesting different kinds of supplies and reinforcement, choosing from a range of predetermined signals like “transports”, “warships”, “money”.
By the second century BC, a new method had appeared. We only know about it via Polybius, who claimed to have improved on an even older method that he attributed to a Cleoxenus and a Democleitus. The system that Polybius described allowed for the spelling out of more specific, detailed messages. It used ten torches, with five on the left and five on the right. The number of torches raised on the left indicated which row to consult on a pre-agreed tablet of letters, while the number of torches raised on the right indicated the column. The method used a lot of torches, which would have to be quite spread out to remain distinct over very long distances. So it must have been quite labour-intensive. But, crucially, it allowed for messages to be spelled out letter by letter, and quickly.
Three centuries later, the Romans were seemingly still using a much faster and simpler version of Polybius’s system, almost verging on a Morse-like code. The signalling area now had a left, right, and middle. But instead of signalling a letter by showing a certain number of torches in each field all at once, the senders waved the torches a certain number of times — up to eight times in each field, thereby dividing the alphabet into three chunks. One wave on the left thus signalled an A, twice on the left a B, once in the middle an I, twice in the middle a K, and so on.
By the height of the Roman Empire, fire signals had thus been adapted to rapidly transmit complex messages over long distances. But in the centuries that followed, these more sophisticated techniques seem to have disappeared. The technology appears to have regressed.
Beacon systems throughout western Eurasia appear to have reverted to the simple, single-signal type, with only minor variations. Beacon chains running from Damascus to Cairo were used by the Arabs of the tenth century, and in the thirteenth century the Venetians had a chain running from Euboea in south-eastern Greece all the way to Venice.They were frequently also used by the Crusader states of the eastern Mediterranean. By the fourteenth century, English chains of coastal beacons warned of attacks by the French and the Scots, and in 1455 a Scottish chain warned of invasion by the English. As we’ve already seen, the English were still using the same basic beacon chain in 1588. These systems were seemingly only improved by occasionally increasing the size or number of pyres — the 1455 Scottish chain, for example, had a scale of one to four bales to be set alight to indicate the imminence and severity of the emergency.
Why were the more sophisticated fire or smoke telegraphs forgotten? The last complex version before the seventeenth century was a Byzantine beacon chain devised by Leo the Mathematician in the ninth century. Leo's chain sent different pre-determined messages a whopping 720-kilometres from the mountain pass protecting south-eastern Anatolia (modern-day Turkey) all the way to Constantinople. With two water clocks at each end of the chain to help the stations keep their time, the sender chose the hour or hours of the day that corresponded with a particular message. (It’s unclear exactly how this worked, but the theory I find persuasive is that there were only four or so possible messages, with for example a single message like “invasion” corresponding to hours 1, 5, and 9 of the day, so as not to have to wait an entire day or more before being able to send the right message.) Using hours limited the choice of messages that could be sent, but it ensured that things did not get too muddled in the 30 to 40 minutes that it would have taken for the entire chain to be lit. With such a massive distance to contend with, and with so many relay stations along the route, there was no possibility of doing anything more complex and prone to error like waving torches. Only a system using giant pyres would do.
The design of Leo’s Byzantine chain hints at one of the challenges with doing anything more complex with such signals: that what could be gained in variety or specificity would be lost in distance. Leo’s system and the most basic single-signal chains could have relay stations spaced over 100km apart over open country. But doing anything more complex by using torches would be indistinguishable at such a distance. Torches spaced about a metre apart could be viewed distinctly just under a kilometre away,so in practice the ancient Roman relay stations along border areas tended to be placed 1.5 to 2.5km apart — presumably with the signalling torches having to be spaced about two or three metres apart. Reduced the telegraph communications to such short distances was not ideal, but it at least provided greater reliability in case of poor weather and prevented the signals being hijacked by enemies setting fires too close to the stations, like those that fooled the Thebans.
One potential explanation for the lack of any further development after the ninth century is thus that the cost of maintaining sophisticated long-distance telegraph systems was simply too great, especially for the fractured and much poorer states that replaced the Roman Empire. Leo’s Byzantine system was itself seemingly either shortened or disbanded entirely after just a few decades, when the threat of Arab invasion temporarily subsided.
But this state-based explanation doesn’t stack up for me. For a start, such signals would still have been very useful for military and other purposes over much shorter distances. They clearly were to the ancient Greeks, and not just to the Romans at their height. Indeed, the basic beacon systems used on the English coastline only tended to be about 20km apart, and could feasibly have been put much closer together. Plenty of states in the nine or so centuries after Leo could also afford to create official postal networks, which required all the additional expense of providing roof, food and fresh water for dozens of fresh horses and riders at every interval. At its height, the Roman Empire’s official postal network typically had relay stations only every 15 to 40 km,sending messages at a speed of about 75 km per day, perhaps stretching to 150 km in a dire emergency. Even the American Pony Express of the 1860s managed a speed of 150 km per day, with relay stations every 15 km.
So there would surely have been plenty of situations before the eighteenth century in which sophisticated telegraphing systems would have been affordable to states. Indeed, we know that there were.
Waving the Flag
Fires may have been ideal for sending messages for distances over a kilometre, but there were plenty of situations in which telegraphing made sense over just a few hundred metres — something that might be done by simply waving flags. There are plenty of straightforward examples of flags being used on both land and sea: the simple raising of a flag to signal an army to attack, to signal the sighting of an enemy, or to command the captains of a fleet to assemble on the admiral’s ship.The ancient Greek or Roman signals for an attack apparently involved raising either a red flag or waving a black flag on a spear. What we’re interested in here, however, is the use of flags to send more varied and complex messages.
In the Byzantine navy of the ninth century, the admiral’s ship used flags to transmit a whole variety of battle orders to the rest of the fleet— crucial when voice and trumpet would be drowned out by “the shouting, the confusion, the roaring of the sea, the rest of the din caused by the collision and movement of oars on the galleys and, much more, the cries of the combatants.” Indeed, Byzantine admirals were advised to wave the flags themselves, so as not to lose any time relaying commands. We know of at least eight specific tactical commands — there were possibly many more — which could be telegraphed by raising the standard, inclining it to left or right, moving it to the right or left of the ship, shaking it, raising or lowering it, or replacing it with flags of different shapes and colours.
Yet this Byzantine system is by far the most complex early flag semaphore system I’ve been able to find. Just like with the more sophisticated beacon systems, it seems as though sophisticated flag semaphore regressed. The only other candidate that I’ve seen mentioned is a system used by the Castilian navy of the fifteenth century. But in this case it seems to have largely involved using torches and trumpets to convey just a handful of pre-agreed messages.There’s no evidence that the Castilian system had any complexity beyond the very basic systems used by the ancients and in the rest of medieval Europe.
Even the English navy of the mid-seventeenth century, when it was rapidly rising to global pre-eminence, used only a basic semaphore system that had not gone very far beyond the Byzantines. Despite even having the added range provided by the telescope, invented in 1608, English signal flags by the 1670s were generally still only used to specify which squadron captains were to come aboard the admiral’s ship so as to then receive their orders in person. Only a handful of other very general commands were telegraphed in real-time, mainly by firing one or two guns while loosing certain sails.It took almost another century until in the 1740s French innovators like Bertrand-François Mahé de la Bourdonnais began to develop a flag semaphore system with more flexibility and specificity — on a principle very similar to that of Polybius, but almost two thousand years later. Only in the eighteenth century did the English naval flag vocabulary begin to really expand. It thus took well over century for naval flag signalling techniques to even catch up with the possibilities of the telescope.
Through the Looking Glass?
Many histories of the telegraph credit the telescope with truly unlocking semaphore’s possibilities. Telescopes certainly made the details and positions of flags and other signs more distinguishable over greater distances, extending their effective range. Robert Hooke, when proposing a telegraph system of his own to the Royal Society in 1684, noted that “we must be beholden to a late invention, which we do not find any of the Ancients knew; that is, the eye must be assisted with telescopes”. Hooke’s proposal was to use an alphabet of hung wooden signs, which in order to be distinguishable at a point 10km away would require a telescope at least 2 metres long.
But was the earlier lack of telescopes the key constraint? I’m not convinced. The telescope does nothing, as we’ve seen, to explain why the naval flag semaphores remained quite so unsophisticated for so long. Nor can it explain even the delay for Hooke’s own suggestion, which appeared almost seven decades after the telescope invention, and even then was not implemented. Indeed, Hooke’s contemporaries, like the natural philosopher John Wilkins, seem to have had no doubt that the older systems could be improved without the need for telescopes too. Writing in the 1640s, Wilkins suggested some simplifications to Polybius’s torch-using system, only as an aside adding that it “may further be advantaged by the use of Galileo’s perspective” — that is, the telescope.As with Hooke, it does not seem as though his suggestions were ever implemented.
Besides, Hooke’s system could also have at least partially mitigated the lack of telescopes by simply making the signs even larger. The first widely-used mechanical telegraph system, implemented by Claude Chappe in France in the 1790s, used lever- and pulley-driven moveable arms that could be up to 9 metres long. The Chappe system did use telescopes — they allowed for the placement of relay stations as far as 25 km apart, beating the later Pony Express’s 15 — but just as with naval signalling, semaphore did not have to be used over very long distances. There could easily have been plenty of shorter-distance applications where such a more complex system would still have been worthwhile.
In fact, there is at least one such case — one that I have never seen mentioned in any histories of telegraphy, and which I found only by accident when looking at something else entirely last year. In the 1660s a foreign visitor to Cornwall noted how pilchard fishers had built a high scaffold on the coast, from which a watcher could identify the pilchard shoals from the colour of the water. Then, holding signs made with feathers, the watcher directed the fishing boats below as to “how they are to go and to which side, and how they must set their nets — in short all he would want to tell them by mouth. And they know how to understand him, but cannot see any fish until they drag them on land in their nets. These watchers or signallers perform very strange posture and movements with their arms, head, body, etc.”There’s no mention, or even much likelihood, of the Cornish pilchard fishers using telescopes, but some version of the method lasted locally until well into the late nineteenth century.
Flash! A-ah! Signal to the Universe
Let’s suppose that you have so far found none of my arguments convincing. You may still believe that the telescope was important, or that there is evidence still out there and undiscovered of more complex early naval signalling systems, or that the cost of frequent relay stations for complex fire signals really was unaffordable to all states between the fall of the Roman Empire and the 1790s. Even if all of this is so, then we still need to explain why we had to wait so long for a telegraph system using flashing lights.
Lights can quite easily be revealed and then quickly hidden again to create complex signals — just think of the dots and dashes of Morse code. Indeed, shuttered signal lamps have been widely used this way since the 1860s, and are still present on many naval vessels even today. Some forms of these lamps could probably not have been invented earlier, because they depended upon certain 1820s chemical discoveries in creating artificial light — particularly the limelight effect from directing an oxyhydrogen flame at some quicklime. Later shuttered signal lamps burned kerosene or used electric arc lights.
Yet the lack of nineteenth-century chemistry did not preclude all possibilities. It was still possible to create a shuttered signal lamp using just ordinary flames, which would at least be visible during the night-time. In 1616 the painter, surveyor and sundial-maker Franz Kessler created just such a device by hanging a burning torch on a hook inside a barrel lined with lead, to prevent it catching fire. He even suggested using a concave mirror to enhance the beam from the barrel’s open end. Along with a vent to release the smoke, and the use of surveying instruments to direct the beam at the intended recipient, Kessler then pulled on a cord to raise and lower a shutter on the beam, thereby sending his signals.
Kessler even devised a rudimentary kind of Morse-like code. His basic idea, during the daytime, was for a signaller to simply wave a white cloth or piece of paper, to be viewed by someone using a telescope two to three kilometres away. The number of successive waves corresponded to certain commonly-used letters — once wave for D, twice for I, three waves for R, and so on. And he applied the same idea to his shuttered barrel for use during the night-time, substituting the waves of the white cloth for flashes of light. Given the greater visibility of flashing lights, the night-time signalling lamp seems not to have required the telescope the same way.
Again, however, Kessler’s technique does not seem to have become widespread, and indeed seems to have been largely forgotten until the 1860s — and this despite publishing during the heyday of print. It might be that the uses for his device were too restricted, given the beams were only visible at night. But there was yet another bright light source that would work extremely well during the day-time and was already available to everyone well before the nineteenth century: the sun.
Here Comes the Sun
In 1821 another German surveyor, Carl Friedrich Gauss, used two mirrors to create a device that could precisely direct a beam of sunlight at a specific faraway point — a technique that was not at all complicated from the perspective of ancient or medieval optics, and which used materials that had been available for thousands of years. This device, which Gauss termed a heliotrope, became a common fixture for surveyors, and was quickly reinterpreted as a telegraph, or heliograph. By making a heliograph’s beam mirror movable, or by adding a shutter, the beam could be rapidly flashed to transmit a Morse-like code. You could do it right now using a simple shaving mirror.
Heliographs became extremely widespread, even despite the rise of the electric telegraph, and despite their limitations (they could be useless in bad weather). They were used by the British army in the late nineteenth century in India, Afghanistan, and against the Zulus, by both sides in the Boer Wars, and by the US army in its campaigns against the Apache. Typical heliograph signals, even when viewed without the help of telescopes, could be seen up to 80 km away. The kinds with the larger mirrors, of up to 20 cm in diameter, could reportedly even be seen with the naked eye at an astonishing range of 160 km. (In practice, however, the British army preferred to use the smaller and more portable heliographs when on campaign, even if that meant sacrificing range.)They could even be adapted to night-time by reflecting the light of a fire or the moon. Directed moonbeams could reportedly still be seen at a range of almost 20 km with the naked eye.
Although the later heliographs might have had an added benefit of using highly-reflective mirrors made using silver nitrate — a process only invented by Justus Liebig in 1856 — there’s still plenty of evidence to suggest that even heliographs with older-style mirrors, including just polished plates of tin, would still have had a range that would have put Polybius and the Romans to shame. Indeed, it’s not actually even clear whether the new Liebig mirrors were yet being used by the British army’s heliographs by as late as the 1870s, when those astonishing ranges were recorded. The sources mention mirrors made with mercury, rather than silver nitrate.
So sophisticated long-distance telegraphy really was an idea well behind its time. We had the materials, the resources, and the understanding. We probably didn’t even need the telescope, even if it later helped. There were certainly no physical barriers to developing Morse-like codes, as Kessler even proved by getting close to one in the 1610s. Yet it was only in the nineteenth century that shuttered signal lamps and heliographs appeared. Imagine how different the world might have been had it made use of them when it first could have. We might have had all the commercial and cultural effects of a ‘Victorian Internet’, as Tom Standage termed the electric telegraph, centuries or even millennia earlier.
Why did it take so long? As I’ve mentioned in all my other investigations — of the steam engine, John Kay’s flying shuttle, or the thermometer — the main constraint was not of science, materials, or demand. It was simply down to a lack of inventors in general, and an even more acute lack of inventors trying to solve the specific problems of long-distance communication. It was only in the eighteenth and nineteenth centuries that with the sheer increase in the number and concentration of inventors, and with their collective attentions able to range over so many avenues for improvement, that the low-hanging fruit of non-electric telegraphy were finally, finally plucked.
Age of Invention by Anton Howes is a reader-supported publication. To support my work, consider upgrading your subscription.
Thank you to Ezra Abrams for framing this question on twitter, and to Judy Stephenson for bringing it to my attention.
R. J. Forbes, Studies in Ancient Technology, 2nd, revised ed., vol. VI, Heat and Heating, Refrigeration, Light (BRILL, 1966), pp.172-3
Tiffany Earley-Spadoni, ‘Landscapes of Warfare: Intervisibility Analysis of Early Iron and Urartian Fire Beacon Stations (Armenia)’, Journal of Archaeological Science: Reports 3 (2015), pp.22–30
Ward Beers, ‘Fire and Smoke: Ethnographic and Archaeological Evidence for Line-of-Sight Signaling in North America’, in Enduring Curiosity, Generous Service: Papers in Honor of Sheila K. Brewer, ed. E.J. Brown, C.J. Condie, and H.K. Crotty, vol. 40 (Archaeological Society of New Mexico, 2014), pp.23-4
Sotirios Christos Peithis, ‘Tactical and Strategic Communications in Ancient Greece, Fifth Century BC’ (PhD thesis, UCL, 2021), pp.193-202
Russell W. Burns, Communications: An International History of the Formative Years (IET, 2004), p.12
Forbes 1966, pp.178-9.
The original, Greek source from the 3rd century gives its examples in the 24-letter Greek alphabet (alpha to theta on the right, iota to pi in the middle, rho to omega on the left), but I’ve converted it into the 23-letter Classical Latin alphabet given it was supposed to have been used by the Romans. Also Burns, pp.15-16.
Philip Pattenden, The Byzantine Early Warning System, Byzantion 53 (1983), p.270
e.g. Anthony Luttrell, ‘Smoke and Fire Signals at Rhodes: 1449’, in The Military Orders, ed. Peter Edbury, vol. V, Politics and Power (Routledge, 2012).
Ralph Anthony Kaner, ‘The Management of the Mobilization of English Armies: Edward I to Edward III’ (PhD thesis, University of York, 1999), p.17, 107, 114, 116, 126, 189.
For full details, see Pattenden.
e.g. the Byzantine system of the 9th Century: Pattenden, p.269
Lukas Lemcke, ‘Imperial Transportation and Communication from the Third to the Late Fourth Century: The Golden Age of the cursus publicus’ (PhD thesis, University of Waterloo, 2013), p.40
Gerard J. Holzmann and Björn Pehrson, The Early History of Data Networks (Wiley, 1995), p.6
Holzmann and Pehrson, pp.11-12
The Taktika of Leo VI, ed. and tr. George T. Dennis (Dumbarton Oaks, 2010), p.523
It’s described in some sources as being to do with the Castilian army or navy in 1340. Following the references, however, it seems to all come from a document written to the Castilian admiral Fadrique Enriquez de Mendoza, with the usual date seemingly actually a typo of 1430. The original reference with a transcription of the relevant document is here: Juan B. Olaechea Labayen, ‘Un Código Medieval de Señales Marítimas. Evolución del Código Marítimo y sus Afinidades con los Códigos Documentarios, Revista de Archivos, Bibliotecas y Museos, 82, 3, (1979), pp. 437-48.
James, Duke of York, Instructions for the better ordering his Majesties fleet in sayling (1660). For a summary of the history of English signalling, see: Barrie Kent, Signal! A History of Signalling in the Royal Navy, 2nd ed. (Hyden House, 2004), pp.2-5
Robert Hooke, Philosophical Experiments and Observations of the late Eminent Dr Robert Hooke (W. Derham, 1726), pp.142-50
John Wilkins, Mercury, or, The secret and swift messenger shewing, how a man may with privacy and speed communicate his thoughts to a friend at any distance, (John Maynard and Timothy Wilkins, 1641), pp.156-62
‘The Journal of William Schellinks’ Travels in England’, Royal Historical Society Camden Fifth Series 1 (July 1993), p.119
Franz Kessler, Unterschiedliche bißhero mehrern Theils Secreta Oder Verborgene geheime Künste (Hieronimo Gallern & Johan Theodor de Bry, 1616), pp.19-26
A. S. Wynne, ‘“Heliography and Army Signalling Generally”’, Royal United Services Institution. Journal 24, no. 105 (January 1880), p.254-5 — see comments by the director of military telegraphs and signalling during the Anglo-Zulu war.
Samuel Goode, ‘Mance’s Heliograph, Or Sun-Telegraph’, Royal United Services Institution. Journal 19, no. 83 (1 January 1875), pp.533–48. See for example the mention of mercury — a sign of the old-style mirrors — on p.535, and the mention of tin plates on p.537.
Will Storr’s point about the shift in status, so that the prestige of discovery was able to break through the blocks of propriety-through-conformity, seems to apply.
I wonder if demand might play a role, though. The obvious rival to these systems is indeed the horse and rider. In many cases isn’t that a winner in terms of cost per bit of information? The exception seems to be times in war where speed is of the essence, but perhaps that is rare?