Both Olive and the BFI rely on the same recent restoration by the CNC/Gaumont. As with so many of Gance’s other films, the release history of the 1938
J’accuse is rather confusing…
Between Gance showing a 4510m (162 minute) print to his producers in January 1938, the film was reduced (with Gance’s reluctant approval) for a test screening at Villeparisis in mid January 1938, then again (with Gance’s even more reluctant approval) for its exclusive first-run at L’Ermitage cinema in Paris in late January 1938, then again (without any consultation with Gance) for its general release in February 1938 – by which point it was only 3300m (119 minutes). It was shown in the UK in April 1938 with a length of approx. 2860m (103 minutes), and was distributed in the US in late 1939 (as
That They May Live) in a print of approx. 2000m (73 minutes). It reappeared in France in 1946-7 when it was reduced to either 2750m (99 minutes) or 2900m (104 minutes) – the only two sources I could find offer different figures here.
The Gaumont restoration (as issued in the US by Olive and in the UK by the BFI) is 119 minutes (3300m) and represents a version that likely accords with the general release version of 1938 (with one notable exception, discussed below). The Robert A. Harris restoration (as issued in the US on VHS in 1991) is 125 minutes (approx. 3460m) and includes about six minutes of footage missing from the Gaumont print. (There was also a VHS release in France in 1997, which is listed as 90 minutes long [approx. 2600m, presuming 4% SECAM speed-up] – but I’ve not seen this copy.)
The footage missing from Gaumont’s version that is included in Harris’s version consists of some very minor cuts to some scenes, as well as some significant (though still brief) scenes cut in their entirety. The Gaumont print also contains at least one instance of a sequence being put in a different place – I imagine that the film’s distributors did this to distract the viewer from what would otherwise be a shoddy censorial ellipsis right in the middle of a scene. The Harris restoration seems to switch from one source print to another for all the above instances – I can only presume he took his footage from at least two different prints. (Given that the above scenes possess subtitles in a different size and font, I’m guessing that they derive from an earlier export print – 16mm, by the looks of it – which already had subtitles. The description by Kramer and Welsh in their book on Gance implies that the 16mm print of 116 minutes they viewed in the early 1970s contained all the additional fragments mentioned above, but was missing the original ending [see below]. It may be that some export prints of
J’accuse were assembled before the French general release cut was finalized, hence a small amount of footage was preserved in the former while being lost from the latter.)
In comments cited by
Stefan Andersson (see earlier post), Harris suggests that the Gaumont print is “the reissue version” from 1946, but this doesn’t account for its longer overall running time. The only element in the Gaumont version that certainly dates from 1946 is the rolling text that appears at the very end. (It is essentially a repeat of an earlier shot, over which new text has been superimposed – i.e. this is Gance’s footage from 1938 but the distributor’s text from 1946.) Other than this one instance, I would be confident to assert that everything preceeding it derives from 1938. This includes what Harris calls the “happy ending”. I’m not sure if he is implying this ending appears only in the 1946 reissue version, and/or if he thinks it was an imposition by Gance’s distributors. This ending is in fact detailed in Gance’s screenplay, as well as being described in the synopsis issued at the premiere, and cited in contemporary press reviews. In sum, it was always meant to be there in 1938 and it always was there in 1938 – no one had need to add it in 1946.
Just to be clear on this issue of “alternate” endings, here is the original order of events in the finale of
J’accuse – as per Gance’s wishes in 1938:
the dead rise up and cause panic, Diaz is burned at the stake by terrified locals, Flo drops dead beside him, Diaz and Flo join the legions of the dead, the march of the dead halts war, universal peace is declared, the dead return to Verdun, The End.
By comparison, this is how the Harris restoration ends:
the dead rise up and cause panic, Diaz is burned at the stake by terrified locals, Flo drops dead beside him, Diaz and Flo join the legions of the dead, the dead return to Verdun, The End.
By comparison, this is how the Gaumont restoration ends:
the dead rise up and cause panic, the march of the dead halts war, universal peace is declared, the dead return to Verdun, The End.
Though the absence of Diaz’s ultimate fate on the Gaumont print is unfortunate, I think the exclusion of the final sequence on Harris’s restoration is far more narratively/morally significant – after all, it negates the entire purpose of Diaz’s mission and also that of Gance’s film.
If you added the one (final) sequence included on Gaumont’s print to Harris’s (otherwise longer) print, you would get something around 128 minutes (3545m) – still around 34 minutes shorter than Gance’s preferred pre-release cut. Whether any of the latter material survives somewhere… well, who knows. But in the meantime, let us be thankful that
J’accuse is being released at all – and in a version which at least presents us with an approximation of what most French audiences in 1938-9 would have seen in cinemas.
(As per the
Beaver review, the BFI release has significant extras whereas the Olive has none at all.)