Chapter 8

Translation

Florence Hummford took two images of the same section of writing, taken with lights set at opposing angles while the camera remained locked in position, and adjusted their gamma curves to create high contrast images. She then inverted the images, making the bright areas black and the black areas white. She then selected the dark areas of the first image and copied them onto the second image, creating a monochrome set of images of the symbols.

She then had the computer trace the edges, converting them into Bezier curves. In essence, she was creating a font, or type face, of the symbols. Once the computer "understood" the shapes, it could analyze the characters.

Having performed the procedure once, she set up a batch process to repeat it on all 47 remaining photo pairs, and while the computer ran through them, she rolled her chair over to the casting of a section of the wall that had been "lost" in shipment to Berkeley. The casting had been made with plaster colored to resemble the concrete of the so-called Town Hall, and she’d set it up at approximately the angle of the pyramid wall.

The uniformity of the characters was consistent enough that she was certain they were not carved by hand but created through some automated process. Where the same symbol appeared in multiple places, each one was identical to the others. She ran her fingers over the characters, and found herself thinking that there was a certain beauty to their design. There was also a vague familiarity to them that she couldn’t put her finger on.

At 48, Hummford had been studying languages since grade school. Short and trim, with long, dark hair she kept up in a bun, she was fluent in thirteen languages and could swear in more than sixty, though she demonstrated this ability only in the presence of close friends. After earning undergraduate degrees in linguistics and mathematics and a graduate degree in linguistics at Harvard, she went on to teach at Cornell University for seven years before her recruitment by the NSA. There, in Maryland, she met the defense analyst who became her husband. They’d been married for seventeen years and had never found time to have children. Like her husband, she loved her work.

Her Macintosh said, "Oh, Florence," signaling that it had finished the batch run, and she rolled back over to her desk. She first ran a simple count, and found that there were 1,044 individual characters. She then had the program run a comparison to eliminate duplicate characters. The answer came back almost instantly: 81.

She then set up the program to search all available databases for matches, which would take a few minutes. As the computer silently went about its task, she considered what she knew so far.

A language with 81 characters indicated that the characters probably represented diphthongs, or sound combinations, assuming, of course, that the culture that created this alphabet used verbal communication, and that their written communication was its representation.

Written languages with a small number of characters usually were based on phonemes, where each character represents a sound. English and all other Western languages were examples.

Languages with a large number of symbols, such as Chinese, with thousands of characters, were based on pictographs, where each character represents a word. For instance, the symbol for house began as a picture of a house. In the early development of written language, pictographs were an easy and obvious choice, but as the complexity of a language expands, it becomes increasingly unwieldy. In the case of Chinese, for example, it’s relatively easy to learn a few characters, but very difficult to become truly proficient.

Ancient Sumerian cuneiform writing had started as pictography, but evolved to a series of around 150 characters, each representing a syllable derived from the words that the pictographs had represented. It gained flexibility in the process.

Assuming the 81 characters used in the writing on the main structure represented the entire alphabet – probably including some form of punctuation – the language they represented would be more similar to a cross between Western and Sumerian writing than to Chinese.

The characters appeared in columns of nine, with even spacing between them, and it was this vertical alignment that led her to suspect that the language was written from top to bottom – or bottom to top, for that matter. But the characters were also aligned horizontally, albeit with more space between columns than between rows, so it was not impossible that the writing was actually horizontal.

Hummford next checked the distribution of characters in the sample, looking for classic patterns present in nearly all known languages. For instance, the letter "e" is the most commonly-used letter in English, while "z" is the least used. In the case of the sample from Town Hall, the distribution was surprisingly even. The most-used character appeared 32 times, and the least-used appeared five times, but most fell into a range of from eight to fourteen times.

The biggest problem she faced in trying to understand this writing was that if the characters indeed represented diphthongs or syllables combined to form words, she had no way of knowing where one word ended and the next began. All Western languages use spaces between words to define word boundaries, and punctuation to define sentence structure. But a written language could also use punctuation to define words as well as sentence structure.

The problem with word definition via punctuation in this sample was that the most common character appeared only 32 times. With a total of 1,044 characters, that would imply an average word length of roughly 32 characters. The average word length in Western writing, by contrast, runs from about five to seven letters. While an average word length of 32 characters was possible, Hummford doubted that this was the case.

She printed out a complete set of all 81 characters on a single sheet of paper, and leaned back in her chair to study them. There was a beauty to the curved wedges that made up this alphabet, and she was again struck by a sense of vague familiarity. There were characters that resembled Roman, Hebrew, Greek and Cyrillic letters to a certain degree, but none were exact matches unless a highly stylized typeface was assumed.

She tried to think what the shapes reminded her of... Yin and yang symbols? The curved blade of a scimitar? She stared at the sheet for several minutes, free associating, and then it came to her. The curved strokes used in many of the characters looked like claws. Like the claws of a large bird, only perhaps not quite as curved.

Tucking that thought into the back of her mind, she studied the various shapes, looking for any inherent clues as to meaning. One might assume that punctuation symbols would be shorter or smaller than letters, since they are most effective when least obtrusive. The period at the end of a sentence, for example, is not easily confused by the eye with a letter.

The characters in her printout, however, ranged from a single stroke to a maximum of the four strokes of what looked something like a mangled "E". There was a single straight horizontal line that bulged in the middle and tapered to points at either end, and a vertical counterpart, but neither was high in the distribution of characters.

"Oh, Florence," her computer called. She turned to the screen to read the results, which were pretty much as she had expected. The computer had scored 26 "hits," with the highest tolerance for variance of 57% and a mean of 22%. The hits were spread over 13 different alphabets.

In other words: "Eeep! Not a match!"

She converted the alphabet she had derived into a Postscript font that could be distributed to her research team, and then called Gordon Winston.

After brief pleasantries she said, "Gordon, I’ve finished an initial analysis of the writing found on the main structure, and I don’t have anything positive to report. We have a lot of work ahead of us."

"Can you give me the layman’s summary?" he asked.

She recapped her findings for him, finishing with the computer analysis. "With only a third of the characters even similar enough to anything we know of to measure, and even then with high degrees of variance, and considering that the matches were spread over 13 different alphabets, the bottom line is that this is entirely new."

"Alien?"

"I can’t say that for certain. But if it existed on Earth – well, actually, it has, come to think of it – what I mean is that it bears little or no similarity to any human writing we know of, so alien origin has to be considered as a possibility."

"So what’s next?"

"I’d like to send what I’ve got out to my research team, to let them get started. Especially Pons at CIA and Reif at Northwestern, if that’s OK with you."

"It’s fine with me." Gordon knew that both men had security clearances. "Keep me posted."

"Don’t I always?" she responded, and they rang off.

Sliding the printout she’d been holding during the conversation onto the desk, she called her husband to say she’d be late, and then got up and went to fetch a mug of coffee. She expected that long hours would be the norm for her for a while to come.