UNORTHODOX DEVELOPMENT SECTIONS

According to the standard model of sonata-allegro form, after the "exposition" section in which the several themes are presented, comes the "development" section, in which these themes are "developed," that is, altered, fragmented, or forcibly combined, usually with modulations in quick succession. Thus developments are often the most original and audacious sections. Nevertheless, it is more or less taken for granted that what is thus "developed" is the thematic material from the exposition.

There are, however, some movements in which the development section is literally unorthodox -- not because of the way it treats the themes, but because the themes that it treats are not, or not only, the themes from the exposition. There are not many such movements, but those few are not merely random or frivolous.

AN ADDITIONAL NEW THEME IN THE DEVELOPMENT SECTION
This is a violation of the basic idea of sonata-allegro form. The exposition states the themes that the movement is about, so that those themes will be developed in the development section. An additional theme in the development is a contradiction in terms. Such an anomaly retroactively belies the function of the exposition.

Beethoven: Symphony no 3 in E-flat major, opus 55 "Eroica" (1804)
     In the first movement, the exposition states, not the conventional two contrasting themes, but an inordinately large number of themes. It would seem that this is enough material for even the most extensive development.
     The development section begins at m152. After treating several of the themes from the exposition in the usual manner, an entirely new theme is introduced at m284. After several more treatments of exposition themes, the new theme is stated again at m322.
     The recapitulation begins at m398, and never even hints at the additional theme. The closing section begins at m551. Most analysts give this the conventional name "coda." but I suggest that it should be called a second development. First of all, it is too long and complicated to be simply dismissed as a coda; secondly, it contains passages parallel to those in the first development; and thirdly and most significant, at m581 it repeats the additional theme from the first development.

Schubert: Symphony no 9 in C major, D944 "Great" (1826)
     The final movement is in sonata-allegro form. Its development section begins at m385, but not with a theme from the exposition. It begins with an entirely new theme, which quotes the famous theme from the finale of Beethoven's Ninth Symphony, not exactly but unmistakably. The development continues with a treatment of the second theme from the exposition in the usual manner.
     The recapitulation begins at m599, and never even hints at the additional theme. The final section begins at m974. Like that in the "Eroica," is not merely a coda but a second development, and for the same reasons. From m986 on, the second theme from the exposition and the additional theme from the development are juxtaposed, revealing that they are so similar that they might be two versions of the same thing. Schubert ended the symphony by triumphantly showing that he was comparable to Beethoven.

THE INTRODUCTION IN THE DEVELOPMENT SECTION
This is no less a violation of the basic idea of sonata-allegro form. The exposition states the themes that the movement is about, so that those themes will be developed in the development section. If the introductory material is employed in the development, such an anomaly retroactively belies the function of the exposition.

Schubert:
Symphony no 8 in B minor, D759 "Unfinished" (1822)
     The first movement begins with an eight-measure introductory phrase. The exposition begins at m9, and consists of two clearly defined themes.
     The development begins at m111. It is the normal kind of development, but only develops the introductory phrase. There is not the slightest hint of the two themes from the exposition.
     The recapitulation begins at m219. It is an almost exact repetition of the exposition, without the slightest hint of the introduction. The coda begins at m329. It is, like the development, based entirely on the introductory phrase without the slightest hint of the two exposition themes.
     This movement is most unusual; it is probably the first of its kind. The themes occupy the exposition and the recapitulation, but they are totally ignored in the development and the coda, which deal only with the introductory phrase. Since this movement constitutes such a radical departure from the norm, it is even more deplorable that the symphony is unfinished.

Tchaikovsky: Symphony no 2 in C minor "Little Russian" (1872,1880)
     (The measure numbers in the following refer to the revision of 1880, which did not change the allocation of the themes to the several sections of the first movement.)
     The first movement begins with an introduction consisting of the Ukrainian folk song "Down By Mother Volga" with several variations. The exposition begins at m54, and consists of two clearly defined themes, without the slightest hint of the introductory folk song.
     The development begins at m159. It only develops the introductory folk song and the first theme from the exposition, ignoring the second theme.
     The recapitulation begins at m216. It is an almost exact repetition of the exposition, without the slightest hint of the introductory folk song. The coda begins at m299. It is based entirely on the introductory folk song without the slightest hint of the two themes.

CONCLUSION
At least in some cases, the model of sonata-allegro form is -- as Hamlet said -- "more honored in the breach than the observance." This is just another proof of the generalization, that once Beethoven had broken the mold, other composers were emboldened to violate the model, even contradicting its basic intent.