Push Button Bertha is one of the first music experiments realized with a computer, at the end of a research conducted by Martin Klein and Douglas Bolitho.

Computer Music Research in the Fifties – In the early fifties, the idea to compose music through a computer becomes a quite widespread topic all over the world, but particularly in the United States. If it’s true that the first algorithmic composition was the Illiac Suite of Lejaren Hiller, and most important experiments were conducted by Max Mathews at Bell Laboratories for the first music software of the Music N family, it’s equally true that similar experiments were conducted elsewhere in the same years, or even much earlier as the case of CSIRAC in Australia.

For some of these early experiments, the main problem is the lack of reliable documentation which allows us to study the main stages and, especially, all results. Even Lejaren Hiller, and his collaborator Leonard Isaacson, in 1959 writes about some research conducted in the years before his Illiac Suite, for some of which did not have any documentation that can testify the results obtained.[Hiller, Isaacson]

The research of Klein and Bolitho – Very different, fortunately, is the case of research conducted by mathematicians Martin Klein and Douglas Bolitho, authors of a study in which the result was a song titled Push Button Bertha. Why this research is still well known today?

Actually it is not a coincidence, and the reason should be identified in the motivations of an experiment strongly desired by the Burroughs Corporation. The latter in 1956 acquires the Electronic Data Corporation, a California-based company known for having built the Datatron computer who, after the acquisition, was renamed Burroughs 205. However, the reputation of these machines pushed the company managers to find a way to pass to the future generations the recollection of these powerful computers. So, for this, Klein and Bolitho were summoned to programming the Datatron/Burroughs 205 to compose music.

Information Theory – The researchers started from the suggestions offered by Richard Pinkerton who, in 1956, publish on Scientific American a paper entitled Information Theory and Melody, describing the possibility to generate musical tones through a computer and a table of rules from which to select specific data.

Pinkerton’s research did not produce concrete results (no music was realized at the end of a purely theoretical research), but the value of his insights was out of question, as witnessed in the following years by many research based on his model.

The Experiment – Among these, in addition to the well-known Illiac Suite, there was an experiment of Klein and Bolitho. The two researchers developed a program that allowed the computer to generate random numbers and their subsequent acceptance or rejection. The latter operation occurred in according to a set of rules expressed, for computing needs, as arithmetic operations.

Expressed in musical terms, each generated number corresponds to one of the seven notes of a diatonic scale, allowing the use of both flat and sharp alterations. Once you have chosen a note, this was validated or rejected based on criteria of melodic acceptability, prescribed through a group of 6 rules.

The 6 rules was defined after a previous analysis of songs taken from the pop and classical repertoire. First, they took into consideration the American top ten in 1956,[Leavy] to obtain the first three rules:

1. In a piece of pop music can be found from 35 to 60 different notes;
2. Every song has a structure of the type AABA, where A can contain from 18 to 25 notes for 8 bars in total, while B, in the same length, it contains 17 to 35;
3. If 5 consecutive notes move in the same direction, the sixth move in the opposite direction.

This first group of three rules were joined by three others, derived from an analysis of Mozart’s works:

1. The prohibition of jumps higher than the interval of six notes;
2. The first note of the section A usually does not correspond to the second, fourth and fifth note, altered in flat, of the reference scale;
3. Notes altered with a flat are followed by others a tone down, while those with altered with a sharp, by a tone above.

After defining the system of rules, the computer was able to make a selection of notes to be checked using the table of preset rules. At the end of the whole process of generation and selection the result was Push Button Bertha (a song in Tin Pan Alley style), whose title refers to the buttons on the computer Bertha, the name given to Datatron used for experimentation.

The score of Push Button Bertha, with text by Jack Owens. Note the music attributed to Datatron computer.

Lyric – To give greater emphasis to the whole experiment, the Burroughs also involved the well-known singer-songwriter Jack Owens (a former member of the American Society of Composers, Authors and Publishers), to write the lyric which I reproduce below:[Ames]

She’s push button Bertha
sweet machine what a queen calculatin’
dalditatin chik with a click
my push button Bertha not to large,
what a charge
electronic supersonic friend
the end

once she’s operatin’
watch her rock and roll cool
and calculatin’ this gal has no heart or soul
she’s push button Bertha auomation divine
pay the light bill and you’re right
she’s mine, all mine

she’s push button Bertha
sweet machine what a queen calculatin’
dalditatin chik with a click
my push button Bertha not to large,
what a charge
electronic supersonic friend
the end

Bertha’s not demanding never
want’s your dough, always understanding
just flip a switch and she’ll go
she’s push button Bertha automation divine
now hear this she can’t
kiss ten weight oil
makes her loyal
dream machine

Music – But, what was the sound result? Being an experiment finalized to obtain a score in common music notation, you can try to play it on a keyboard or piano, to appreciate the melodic yield. Or, you can try to transcribe the original score in MuseScore and then convert it as an audio file, exactly what I did. The result probably will not be faithful to the original but I think it makes an idea:

While listening you will understand that you have to deal with a simple song with a nice melody. Overall a nice result, and certainly much better than many supposed human composers. Clearly what matters in this case, is not so much the end result but first the idea of ​​making music through a computer, a very incredible fact when viewed through the eyes of a man of the fifties.

The Press Reaction – The experiment sought by Burroughs still hides some other curiosities. According to the memoirs of Lejaren Hiller, the research was done around July 3, 1956; in the same day on page 51 of the New York Times an article was published. It was titled Brain’ Computes New Tune for TV, which formalizes the outcome of the experiment of Klein and Bolitho. [Hiller, Isaacson] Elsewhere I have read that the search was started in the spring.

One of the many articles published in that years. Toledo Blade, July 11, 1956.

Beyond the exact date, though important, it is clear that in the following months and years, many have continued to speak of this research, as demonstrated in the brief press review, but the experiment not was the only  hot topic. In fact there was another discussion: the problems encountered by the Library of Congress for the copyright of a song made ​​by a machine.

After one year, many American newspapers still published the news of the experiment of Push Button Bertha. Reading Eagle, July 4, 1957.

The Television – To emphasize experimentation, Burroughs decided two strategies: first, the execution of Push Button Bertha in a television broadcast. Second, the realization of a recording to be sell to the general public. As for the broadcast there were no particular problems, and there was July 16, 1956 in a scientific broadcast of the KABC-TV in Los Angeles.

Copyrights – Quite different the recording question, that produced some problems. In fact, the U.S. Library of Congress, responsible for defining the copyright, he found himself facing a whole new case and legally virgin. Who is the owner of copyrights of Push Button Bertha? While Jack Owens was undoubtedly the author of lyric, who could acquire the right of authorship of the music? The two mathematicians, or computer Datatron? Or the owner of the Burroughs computer? Well, the issue remained unresolved, and five copies of Push Button Bertha remained unsold because it cannot be sold publicly by the absence of property.

After all, it’s clear that Push Button Bertha was a well-known case in the United States, which still allowed us to be aware of this interesting experiment of computer music. After this, a more in-depth research was conducted by Lejaren Hiller and Leonard Isaacson for the Illiac Suite, but this is another story. Equally interesting is the question of the copyrights, in particular for a work realized in recent years by David Cope, who relived a similar problem in a much more elaborate and unexpected work: that of Emily Howell.


To write this post I’ve read:

Anonymous, Brain’ Computes New Tune for TV, New York Times, July 3:51, 1956.

Charles Ames, Automated Composition in Retrospect: 1956 – 1986, Leonardo, Vol. 20 [2], 1987, pp. 169 – 185.

David Levy, Robots Unlimited: Life in a Virtual Age, A. K. Peters, 2006.

Lejaren Hiller, Leonard Isaacson, Experimental Music: Composition with an Electronic Computer, McGraw-Hill, 1959.

1 Comment
Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.