A series of random numbers could easily be used as an index to derive a series of Fibonacci values. For example, if we had random numbers from 1-9, and used them to index the first nine elements of the classical Fibonacci series, that table would look like so:

This means that if the random number you got was a 6, the Fibonacci value you used for this decision would be an 8. Getting a random 7 means using the Fibonacci value 13.

So there I was, with my two piles of books and my crude algorithm. I selected which pile of books, which book, which pages, and positions on pages using some other kind of random number choice (I don’t remember which process I used for those numbers), and then took a number of words from the chosen text and wrote them down. There would always be only a Fibonacci number of words of text chosen from the book to be written down. I made one collage of words in this way, scratching out mistakes as I went, and when I was done, I made a second collage using, if memory serves me correctly, the same algorithm and the same two piles of books used to make the first collage. Different random numbers, of course, meant that the second text was totally different to the first. These two texts nicely fit, hand-written, onto one page.

I then decided to make two more collages, using the same process, but my two sources for information would be the first two paragraphs I had written, rather than the original two piles of books. I chose a random number between 1 and 2 – this told me which paragraph to derive my text from. Next, I counted the number of words in each paragraph, and chose a random number less than that value – this determined where I would start in the paragraph, and then I would draw a random Fibonacci value number to determine how many words I would take from the chosen spot in the chosen paragraph to write down. I did this process twice, generating two more paragraphs. Again, I hand wrote these two paragraphs on a single large sheet of paper, scratching out mistakes as I went. I did this process five times, resulting in a five-page text with ten paragraphs on it. Since each generation of texts used the previous generation as source material, the vocabulary of each generation became more and more restricted. Even by 5 generations in, it was already clear that this kind of process was occurring. I made a photocopy of this text and gave it to Kathy as a birthday present. She was delighted.

I then showed the text to Kenneth Gaburo at our weekly lesson, and he was also enthusiastic. He had ideas about the text. Looking at the handwriting, he said, ‘Well, obviously, you have to perform this.’ That idea hadn’t occurred to me. But, I agreed. And he then said, ‘and your handwriting is such an intrinsic part of the text that you have to figure out a way to use that as a cue to your kind of performing. And those scratch-outs should be included as part of the performance.’ Since I was also a member, at the time, of the Extended Vocal Techniques Ensemble, which was rehearsing the limits of the human voice, that instruction was easy. A simple (inhaled or exhaled) multiphonic could easily be performed using the shapes of my scratchings-out as a guide to the contour of the multiphonic. In this way, I could read the text in a way that reflected my handwriting and in which I would insert vocal multiphonics (these might sound like screeches of various kinds) at various times, at places determined by the mistakes I had made while writing out the poem.

My interest in mistakes was an essential part of the compositional thinking of a number of us in San Diego at the time. As well, Kenneth suggested that each punctuation mark might indicate some kind of physical gesture that I might make while I was performing. Inspired by the musical comedian Victor Borge, who had done routines where each punctuation mark in his text was realised as a kind of vocal and physical gesture, I incorporated gestures into my performing, with all commas realised the same way, all full stops realised in another way. These, occurring where they did in the text, was yet another level of text-derived structuring that was incorporated into the performance.

I told Kenneth that I had noticed that the texts were having a narrower and narrower vocabulary with each generation, and that I was curious as to how far that process could be extended. He, of course, in the manner of any (good) teacher, encouraged me to follow this process through, no matter how much work that might entail. So, a couple of weeks later, I returned to the text, and texts 9 and 10 became the basis for texts 11 and 12, which became the basis for texts 13 and 14. Eventually, I reached texts 39 and 40, by which time the vocabulary had narrowed down to mostly repetitions of the word ‘Nighthawk’. My handwriting had changed as well … from the loose, relaxed writing of the first poems, it gradually got more and more upright, tight, and manic. The final text was a 20-page text of 40 paragraphs. When we did a live performance run-through, it took about an hour to perform, and over that hour, my performing also became more and more manic, until at the end I was breathlessly and manically performing a mantra of words and gestures centred around the word Nighthawk. This had gradually evolved over the course of the hour of the reading. I was, and still am, exhausted by the end of any performance of this part of Nighthawk.

So what we had was a structure that, although it used the Fibonacci series to generate one level of structural detail (the number of words derived from each text), it was actually generated more by the physical act of writing itself. Each element of the writing – density of words, shape of writing, mistakes and their scratching out, punctuation used in the original texts – was used to shape the overall performance. In the past, I think, artists had used tools like the Fibonacci series to structure their work because they felt it gave their work some kind of inner coherence, and, in an almost mystical sense, reflected proportions of natural phenomena as well.

My use was simply to get an interesting spray of random words, and the use of both the random process and the Fibonacci series was simply a means of giving my arbitrary choices a bit of structure and discipline. Once the materials were generated by a numerical process, the physical act of notation – the handwriting of the text – became the source for all aspects of the performance. So here, all the elements that a number of us were using to make musical compositions became the basis for a performance poem. The performance act made Nighthawk one of the most physical, and physically demanding pieces I had ever done.

I went on to expand this portion of Nighthawk – adding slides derived from colourising a video of me performing it, and later wrote two more works, Nighthawk II and Nighthawk III: Bittern, using similar processes and expanding the ideas into making collages of found sounds, electronic sounds, vocal sounds and video synthesis. From 1976–1979, I did about 14 performances of the full three and a half hour cycle of three pieces in Australia, USA and Europe, as well as doing many more performances of just the first part of Nighthawk as a solo reading. Additionally, other composers, such as Ned Sublette and William Brooks, took the 40 texts of Nighthawk and gave each text to a different performer who made their own unique realisation of it.

The hand-written text was later published by Lingua Press – Kenneth Gaburo’s publishing house –and after his death in 1993, it was acquired by Frog Peak Music (where a few copies are still available). Somewhere in my piles of stuff, I still have about 20 or 30 copies of the score, but at the present time, I can’t seem to be able to find them. No worry. They’ll turn up eventually.

Because handwriting was an essential nature of the piece, I have resisted any notions of typing the text up. Scans of a couple of the pages of the score are included here to give some idea of how the score looks.