On June 18, 1908, 781 people came ashore at pier number 14 in the port of Santos—and not one of them, as far as we know, went to the telegraph office to tap out a message home in Japanese. But if one of them had tried, it would have made for a story in which script, electricity, and bureaucracy collide—and none of them yields.
There is no reliable evidence that any specific passengers aboard the Kasato Maru sent a telegram home using the Wabun code. No such episode appears in diaries, correspondence, or the archives of Brazilian telegraph companies. Everything that follows is a counterfactual scene. A thought experiment that lets us examine the real communications infrastructure of an era when the first major organized wave of Japanese immigration to Brazil was only beginning. It seems to me that events that never happened sometimes reveal the workings of the world best: not what did happen, but what might have happened and why it did not. The ship left Kobe on April 28, 1908, spent fifty-one days at sea, and brought 600 men and 181 women, plus 12 independent emigrants—people from Okinawa (325 people), Kagoshima (172), Kumamoto (78), Fukushima (77), and Hiroshima (42). This was the beginning of an organized flow, a key symbol, not the first contact between the two countries as such—diplomatic and commercial threads had stretched between them earlier. But it is from that pier in Santos that the history of Brazil’s Japanese community begins—the largest outside Japan itself.
The phrase “send kanji by Morse” sounds specific, but it is wrong—much like “send a painting by telephone.” The Japanese telegraph code Wabun did not encode kanji. Thousands of logographic characters—each with its own meaning, often with several readings—could not fit into a binary system of dots and dashes. The code worked with katakana: each syllabic character—there are several dozen of them—was assigned a sequence of short and long signals. A text written in kanji first had to be turned by the operator into a phonetic transcription—broken down into syllables, as if he were dictating the message aloud, except he was recording not sound but electrical impulses.
This transformation was no small matter. Japanese is saturated with homonyms: dozens of words can sound the same while being written with entirely different kanji and carrying entirely different meanings. Is the word kōen a park? A lecture? Support? A performance? On paper, a character resolves the ambiguity instantly. In telegraphic kana, the context vanishes, and the recipient at the other end of the wire has to guess. Proper names were a separate headache: a Japanese name written in kana can be read several ways, and even a native speaker is not always sure which version was intended. Addresses, numbers, service notes—all of it required separate agreements, special signs, additional symbols that made an already cumbersome transmission even heavier.
In essence, the telegraph forced a language with thousands of visual signs through the eye of a needle: a set of several dozen syllabic characters, each one a chain of dots and dashes. A writing system that had relied for centuries on visual distinctions of meaning found itself in a world with only two states—“signal” and “pause.”
The history of Wabun does not begin with the Japanese. The first Japanese telegraph code was created by the Dutch in 1854, and on July 2, 1855, it was demonstrated in the Hamarikyu gardens before shogun Tokugawa Iesada. Katsu Kaishu and Oda Matazo were present at the demonstration—men standing at the crossroads of old and new Japan. The shogun watched words turn into clicks, and clicks turn back into words. It was not a cipher; there was no secrecy—only a way of representing text, translating it from one system of writing into another.
In 1869, Koyasu Shun proposed his own version of the code, followed in 1873 by Yoshida Masahide and Terasaki Son, drawing on the Dutch DÖTV 1854 table. They used katakana and the traditional iroha order—an ancient poem in which every syllable of the Japanese syllabary appears exactly once, a kind of Japanese “ABC.” Frequency optimization—assigning the shortest codes to the most common letters, as Morse did for the English E (a single dot)—was not used. The code was uniformly inefficient, if you like: democratically slow for every syllable.
By 1904, a new approach had appeared: each kana character was encoded as a two-digit number, the numbers combined into five-digit groups for international transmission. From May 1, 1905, Foreign Minister Komura Jutaro introduced a modification—all numbers were increased by one. Why? To avoid confusion with zeros and service combinations in international telegraph traffic. One added unit—and the entire code shifts, like a gear turned by one tooth. I am fascinated that the fate of an entire language in the wires depended on such arithmetic trifles.
Let us suppose one of the settlers from the Kasato Maru did make it to the telegraph office in Santos. The office existed: Western Telegraph opened one at 14 Praça Mauá in 1906, two years before the ship’s arrival. The physical possibility of sending a telegram from Brazil existed. But where would the signal have gone?
A submarine cable had linked Lisbon and Recife since June 23, 1874—a transatlantic thread through which South America reached Europe. From Recife, the signal could travel down the Brazilian coast to Santos, then across the Atlantic to Portugal, onward along European land lines eastward, through Russia to the Pacific coast. The Vladivostok–Nagasaki cable, laid as early as 1871, connected the continental telegraph network to Japan. A line from Nagasaki to Tokyo had been operating since 1873. There was also a Pacific route: in 1906, the Commercial Pacific Cable Company linked Guam to Japan via the Bonin Islands, closing the trans-Pacific ring. Roughly speaking, by 1908 the world was laced densely enough with wires for an electrical impulse to get from Santos to Kobe—even if it had to pass through half a dozen relay stations.
But a physical connection and the ability to transmit Japanese text are different things. At every intermediate station sat an operator who received the signal, wrote it down, and passed it on. The Brazilian operator in Santos knew International Morse code—Latin letters and numerals. So did the Portuguese operator in Lisbon. The Russian operator in Vladivostok probably worked with both Cyrillic and Latin letters. But Wabun? A code tied to katakana, to the iroha order, to five-digit numerical groups? Only a Japanese operator at a Japanese station understood it.
The St. Petersburg Convention of 1875—the main set of rules governing international telegraph traffic—required telegrams in “plain language” to be composed in a language understood in the contracting states, or in Latin. For extra-European communications, a word was limited to ten characters, and the franc was the unit of account. Japanese text written in kana was not “understood in the contracting states”—no Brazilian, Portuguese, or Russian operator could read it. Numerical groups—five-digit numbers into which kana syllables were packed—formally counted as an “encrypted” telegram, and encrypted telegrams cost more and were not accepted on every line.
This is probably what the route of the imaginary telegram would have looked like. The settler comes into the Western Telegraph office and dictates the text. The operator does not know Japanese. The settler must transliterate the message himself—either into Latin letters (rōmaji) or into numerical groups. If in Latin letters, the text goes out as an ordinary international telegram, but at the receiving end in Japan it must be turned back into kana or kanji. If in numerical groups, the text travels as a coded telegram—more expensive, slower—and at every intermediate station the operator handles it as a meaningless string of numbers, with no way to check for an error. One transposed digit—and the syllable changes, the word comes apart, the meaning slips away.
Cost pressed hard too. An international telegram from Brazil to Japan in 1908 meant thousands of kilometers of cable, dozens of retransmissions, payment for every word. The settlers aboard the Kasato Maru had come to work on coffee plantations in São Paulo state, many of them under contract, carrying debts for their passage. A telegram across half the world would probably have cost more than a week’s wages. A letter aboard a steamship took months, but it was almost free. Economics settled the question before encoding ever could.
What gets me most in this story is this: a problem we are used to thinking of as a computer problem—the problem of encoding, script compatibility, transliteration—arose half a century before the first computers. The telegraph was the first system to demand that human language be turned into a linear sequence of discrete signals. For English, French, German, that was relatively simple: an alphabet of several dozen letters, each one a unique combination of dots and dashes. For Japanese—a scaling catastrophe. Thousands of kanji would fit into no reasonable code, and the language had to be “lowered” to the phonetic level, losing along the way the visual unambiguity built into the writing system itself.
It is exactly the same task that the creators of Unicode would tackle a hundred years later: how to fit all the world’s writing systems into a single code space without losing meaning. Nineteenth-century telegraph operators solved it by hand—a Dutch engineer before the shogun, a Japanese official with an iroha table, a Brazilian operator peering at unfamiliar numbers on a tape. Each of them was a living converter, a human interface between two systems of information representation.
The imaginary telegram of a settler from the Kasato Maru is not a story about the telegraph. It is a story about what happens when language encounters a communication channel narrower than the language itself. The channel dictates the rules: break your characters into syllables, the syllables into numbers, the numbers into dots and dashes. At the other end of the wire, someone will assemble the dots back into dashes, the dashes into numbers, the numbers into syllables, the syllables into characters. If luck holds and not a single operator at half a dozen relay stations makes a mistake in a single digit. If the recipient in Kobe or Kagoshima guesses which of five homonyms was meant. If the sender has enough money.
We do not know whether any of those 781 people sent even one telegram from Santos in June 1908—by any language, in any code. Most likely, they did not. But the very possibility—and impossibility—of that telegram tells us more about the world of 1908 than many a dispatch that was actually sent.