In addition to the useful and correct comments above, I'd like to add that the registrations of Franck's "Chorals" are probably not by the composer himself. Nevertheless, they are very appropriate, but with this knowledge, the performer can afford some more liberties...
This is an interesting point. The series of manuscripts stored in the Bibliothèque Nationale (MS 8707) give many schemes by Franck, for various types of registration - a fair proportion of which are clearly identifiable as referring to his own instrument in Ste. Clothilde, and others, that in the Trocadéro. The manuscript for the Premièr Choral is, I believe, the property of The Pierpoint Morgan Library, NYC. The description given is thus: 'Autograph manuscript, unsigned, dated 7 August 1890. Thirteen pages of music. Dedicated to Mlle Clotilde Bréal. No registration indications. (The manuscript was acquired by Alfred Cortot, sold after his death in 1962 to Robert Owen Lehman, and given to The Pierpoint Morgan Library in 1972.)' 1 However, the engraver's plates for the Premièr Choral and the Troisième Choral are missing, apparently - as are early drafts of these compositions.
Notwithstanding, the registration directions (and it is always better to read these in French and ignore any English 'translations') are appropriate in style (with the caveat supplied by Vox Humana) - and to the best of my knowledge, in the case of the Trois Chorals, are entirely possible to realise on the original instrument in Ste. Clothilde. However, as Vox says, this is really only half of the story. Some years ago, I was with a colleague at the organ of his church, where he was demonstrating the (large Romantic) instrument's capabilities. At one point, he said "And this is my French channel" - and then proceeded to draw all of the 8ft. foundation stops, including the huge (and probably leathered) Open Diapason I, the fat Hohl Flute, and a number of other unsuitable ranks. Unfortunately, he had attempted to follow the letter of the 'law' (as it were) - rather than the spirit.
When playing either the First or Third Choral at the Minster (J.W. Walker, neo-Classical with eclectic elements, 1965), I tend to eschew virtually all of the compound stops, except the G.O. four-rank quint Mixture, which I set to draw with the reeds. Otherwise, I rarely draw anything above 4ft. pitch - other than the Swell Fifteenth, which helps to brighten the rather 'English' 16ft. and 8ft. chorus reeds. Fortunately, the Swell Clarion is rather brighter, and thus is tremendously useful when playing much French Romantic music.
As Sprondel states, the Céleste should never be used unless specifically requested. In the whole of Franck's œuvre, I do not think that he called for this stop more than twice. It is worth remembering that the Hautbois (or Basson-Hautbois) was considered to be part of the 'Fonds'. In larger instruments by Cavaillé-Coll (i.e.: those with divided layes, or 'chests'), this stop was almost invariably sited with the Jeux de Fonds, as opposed to the Jeux de Combinaisons (reeds and upper-work).
1 p. 145; French Organ Music from the Revolution to Franck and Widor; Edited by Lawrence Archibold and William J. Peterson. University of Rochester Press (1995).