Versions and censorship of Eyes Wide Shut

A copy of Eyes Wide Shut may differ in what it reveals without containing a different sequence of events. The history of its versions requires us to distinguish visual intervention, editing, classification and technical presentation.

The US alteration changed what was visible

In his 1999 release review, Roger Ebert described the addition of digital figures that obscured parts of Somerton’s sexual activity to secure an R rating in the United States. This intervention affected what could be seen within certain shots. It did not necessarily involve removing an entire scene or replacing the whole episode with a different narrative sequence.

The distinction matters because obscuring something with an added body changes the composition. An element that appears to be in the foreground can prevent a part of the background from being seen. The action behind it remains less accessible to the viewer, even though the shot retains its duration. The analysis of Somerton must identify which version it refers to when discussing these relationships of looking, rather than treating all copies as visually identical.

The term international version identifies a presentation without that US digital obscuring. Criterion explicitly describes its restoration, supervised by Larry Smith, in those terms. That label does not prove the existence of a lengthy secret version or a narrative expansion. It indicates a specific difference from a distribution intervention, which should be described in terms of its method and scope, not through rumours about missing minutes.

The study of masks and appearance examines how the film distributes covered and exposed bodies within the fiction. The added figures belong to a different level: an alteration to the image intended to affect its exhibition. Confusing these two levels would mean attributing to the original staging a decision that a comparison of versions should examine as a material intervention in certain frames.

Classification and running time describe different things

The BBFC records the British cinema copy classified on 1 September 1999 as 18, with a running time of 158 minutes and 47 seconds. Its catalogue also lists a Warner home-video edition classified on 11 April 2000, with a running time of 152 minutes and 26 seconds. These are records of identified items and dates. The numerical difference alone does not prove that six minutes of events were removed.

Playback speed can explain substantial variations without changing the number of frames. If material shot at 24 frames per second is played at 25, it takes less time. Applied as a calculation to 158 minutes and 47 seconds, this gives approximately 152 minutes and 26 seconds. The correspondence is consistent with that difference in speed; it does not replace a shot-by-shot comparison verifying every feature of two specific copies.

A comparison between two editions needs to align the events, not just the player counters. If the speed changes, the same minute may correspond to different actions. It is worth locating a cut or a recognisable gesture and observing from there which elements appear within the frame. It is also necessary to distinguish a figure obscuring the background from an area excluded by the image crop. These are independent differences: adjusting the size of the playback windows does not remove an obscuration incorporated into the shot. This procedure makes it possible to describe what changes without attributing every discrepancy to a lost scene.

Age classification addresses a different issue. A British category does not mechanically translate into its supposed US equivalent, because the institutions and their procedures are different. Nor does it, by itself, tell us anything about the picture format or sound track. A precise description records the country, body, date and medium before comparing labels that might seem similar to the reader.

The production schedule should not be reconstructed from these running times. The duration of a file, the total time taken by the credits and the logos included can vary without revealing a new narrative decision. Before discussing deleted scenes, a difference in content needs to be identified; simply subtracting the minutes listed in two commercial or administrative records is not enough.

Framing, discarded material and released versions

The visual presentation has also changed between editions. Todd Doogan documented the US Warner DVD from 2000 with a 1.33:1 frame; Criterion identifies its restoration as 1.85:1. These are different proportions of the visible area. They do not automatically amount to a different story, nor do they allow us to infer a single intention for every format. To analyse a specific change, the same shot would need to be compared under equivalent playback conditions.

A taller image may show areas excluded from another presentation, while a transfer or restoration can alter detail, colour or texture. Without evidence, none of these operations should be confused with the inclusion of additional footage. The history of the copies brings together different decisions that can coexist: framing, resolution, mix, subtitles and the treatment of particular shots. Describing them separately avoids using the term director’s cut as a universal explanation.

It is also worth distinguishing a written text, a filmed scene, material discarded during editing and a version that was actually released. The existence of a proposal in a draft does not prove that it was filmed. The study of the Kubrick archive explains why documentary identification matters. A fragment absent from the release needs a verifiable provenance before it can be treated as evidence of a later suppression.

The analysis of the final cut and editing authorship examines work carried out after the director’s death, while translations and dubbing call for a different kind of comparison. The differences documented here support a more limited conclusion: there are identifiable changes to visibility and technical presentation. They do not justify filling in the uncertainties of the fiction with a hidden film whose existence has not been demonstrated.

References: Roger Ebert, Eyes Wide Shut, release review, 1999. British Board of Film Classification, Eyes Wide Shut records dated 1 September 1999 and 11 April 2000. Todd Doogan, Eyes Wide Shut (all versions), The Digital Bits, US Warner DVD from 2000. Criterion Collection, listing for the restoration of the international version of Eyes Wide Shut. The running-time calculation compares playback at 24 and 25 frames per second and does not replace a comparison of the copies.

View all articles in Eyes Wide Shut (1999): a guide to the film