A Numerate Film History? Cinemetrics Looks at Griffith, Sennett, and Chaplin (1909–17) (review)

Encounters between century-old films and today’s computational statistics challenge received understandings of film history, how films are analyzed, and the history of film studies. Tools of the digital age, such as the measurement software Cinemetrics, make it easy to count and measure the length of shots, calculate average shot length (ASL), and catalog the cinematographic qualities of shots. More than just a tool for local, one-time usage, Cinemetrics is an online collaborative project that allows scholars to collect, share, and compare statistical data on films. The value of and reservations about the application of such data in the study of moving images were the focus of the one-day conference “A Numerate Film History?” at the University of Chicago, March 1, 2014.¹

At the center of discussion was the applicability of statistics for the study of stylistic patterns in film history, both their continuity and change. Participants raised concerns about the values that numerical data represent and how the data are interpreted. Such debates expose the tension between the perceived objectivity of numbers and the bread-and-butter subjective interpretation more typical of traditional humanities. The conference organizers, responding to such skepticism, promoted the use of statistical data as an approach to studying the poetics of cinema by demonstrating how statistical data could be used to corroborate and challenge accepted facts of film history and theory.

One of the aims of the conference was to establish a historical context showing how numbers and measuring techniques have figured in the history of film studies. In his introductory remarks, Yuri Tsivian (University of Chicago), who developed the Cinemetrics tool, cited the research of Reverend Dr. Stockton as a predecessor of modern scholars who time film shots. In 1912, Stockton counted the number of shots, intertitles, and inserts in films released by different studios of that year. His measurements and conclusions were reported in Moving Picture World for August 10, 1912.² Stockton anticipated modern stylistic scholarship as well as historical attitudes regarding cutting rates. Long takes for him were a prerequisite for drama. He therefore was appalled by D. W. Griffith’s The Sands of Dee (1912), which contained seventy-five shots squeezed into fifteen minutes.

In his keynote address following Tsivian’s initial remarks, Tom Gunning (University of Chicago) traced the liaisons between numbers and moving images to the very beginning of cinema. The first motion picture to be copyrighted in the United States, Edison Kinetoscopic Record of a Sneeze (1894), was likely never shown as a film but was presented as a page of enumerable, discrete images. Along with the chronophotography of Eadweard Muybridge and Étienne-Jules Marey, Edison Kinetoscopic Record of a Sneeze represents how numerical values figure in the origins of moving images.

During his presentation, Gunning insisted on the difference between considering films through length of shots and editing patterns as opposed to the films’ content and perception of them during projection. For Gunning, there is a troubling disparity between numbers and discrete images, on one hand, and the continuity of watching and perceiving a film, on the other. Such concerns pose important questions for film scholars who make use of statistical data. Does looking at the ASL of films involve a division between the discrete images and the experience of watching these images in motion? Does the use of statistics in film studies imply a false objectivity to supplant the subjective nature of interpretation? Last, can statistically minded scholars correlate editing patterns with concepts of narrative structure without confusing mathematically derived, statistical graphs for diagrams of plot structure that chart parts of the narrative structure, such as exposition, rising action, crisis, climax, and denouement, on a triangular or cresting shape? After Gunning raised these questions, the subsequent three presentations responded to this skepticism and demonstrated the value of using numerals and quantitative data for film analysis from the perspectives of statistics and film studies.

Michael Baxter, Emeritus Professor of...