Michael Noll & Kenneth Knowlton

Eliza Pertigkiozoglou
4 min readApr 30, 2017

Bell Telephone Laboratories and Early Computer Graphics

pages from Design Quarterly 66/67, Bell Telephone Labs computer graphics research. Left: Noll’s approach, Right: Knowlton’s approach.

During the 1960s, creative research was conducted at Bell Telephone Laboratories, with a significant number of contributions in the area of computer graphics. Revisiting these early achievements is informative not only regarding the origins of current forms of digital image formats but also for understanding the associations of elements of work and sequence of operations implied by the computer when used for visual representations. In 1966, in the issue of Design Quarterly magazine dedicated to “Design and the Computer”, the engineers Michael Noll and Kenneth Knowlton from Bell Labs explain two distinct ways of coding design into graphic form. Both derive from the acceptance that the computer can work with a predetermined set of instructions and has great computational capabilities but use different strategy to describe the visual image to the machine.

cover of Design Quarterly 66/67 issue dedicated to Design and the Computer

Michael Noll, engineer and researcher at Bell Labs for 25 years, was one of the first so-called “computer artists”. His first two-dimensional computer art “Gaussian-Quadratic”, produced in 1962, was exhibited at Howard Wise Gallery in New York in 1965 along with the European Computer artists George Nees and Friedrich Nake, with whom he was in close dialog. Noll was arguing that computer could play an important role in linking art and science, with both sides learning from each other. According to him, shapes and motions important to the designer as an artistic medium have a lot of things to share with techniques for visual display of scientific data, such as motion, rotation of n-dimensional objects etc.

source: Design Quarterly 66/67, top: image generated by random probability, bottom: Noll’s artwork “Gaussian-Quadratic” with Gaussian probability generator

In the article, he explains the process of production of a two-dimensional image. Based on the fact that the computer can manipulate numbers based on given instructions (program), he suggested commands for drawing straight lines between numerically specified points. The program creates arrays of points in geometric relationships that are later assigned values (i.e. point coordinates) and are connected with each other. The process is repeated to a given frame rate. A generator of a random sequence creates “interesting” randomness to the picture (a sequence of numbers would be described as random if an observer were unable to determine a formula that produced it). This sequence falls between two limits and controls the occurrence of a number based on a probability density. Since total randomness does not lead to compelling results, Noll suggests randomness with mathematical order and in particular the Gaussian probability density in which the occurrence of a number is greater close to the average. Noll works with geometric elements of pictures (points, lines etc.) and their analytical description.

Noll’s example of drawing a square:Four points are located, in a quadrant of the X-Y coordinate system. An instruction to the microfilm plotter to connect these points and thus to produce a square. source: Design Quarterly 66/67

Contrasted to the line drawings of Noll, which are composed of hundreds of line segments, his friend and colleague Dr. Kenneth C.Knowlton, was working on “mosaic-type” pictures consisted of thousands of spots for tiny characters. His computer language BEFLIX contains instructions for automatic production of pictures and motion graphics. His argument was that, instead of specifying pictures in terms of their elementary components, such as spots and lines, he proposes to talk in the language of the machine. “Type such and such a title. Center each line, give the letter shadows, and then shoot 150 frames”. Knowlton sets a grid of elements as the main unit of operation and assigns a scale from 0 to 7, which indicates the intensity of light at that element. For example, the instruction for drawing a straight line, in this case, requires the programmer to specify beginning and end points and then the intensity of the shade of gray. By this way, he reduces the very large number of corresponding instructions for the microfilm recorder to just a few statements.

Frame from BEFLIX-produced film. This narrated film explains the working of a new computer language by means of moving “bugs”, block diagrams and letters. source: Design Quarterly 66/67
Frames from computer-produced film “A Computer Technique for the Production of Animated Movies” produced by programming in the BEFLIX language. source: Design Quarterly 66/67

Both of these early experiments in computer graphics hint at their descendants, vector and bitmap images respectively. Additionally, the way these pioneers described their workflows and assumptions regarding the capabilities of machine language and programmability in relation to the generation of graphic displays gives an insight on two distinct ways of synthesizing and manipulating visual content on the computer. Noll’s approach manifests topological associations between distinct elements, which conceptually belong to the family of graphs. Knowlton, on the other hand, with his insistence to communicate with the computer in a way that matches its own language, creates modules of elements — placeholders- that are assigned on-and –off or intermediate values.



Eliza Pertigkiozoglou

Architect, Computational Designer, PhD student at McGill University