Why does it matter?
1. Have a look at KnT 1.2676-83:
This fierse Arcite hath of his helm ydon,
And on a courser, for to shewe his face,
He priketh endelong the large place
Lokynge upward upon this Emelye;
And she agayn hym caste a freendlich ye
(For wommen, as to speken in comune, 2681
Thei folwen alle the favour of Fortune)
And was al his chiere, as in his herte.
The couplet in parentheses (2681-82) is omitted in thirteen MSS, including some of the best MSS (e.g., El, Hg, Ad3, Gg). According to Manly and Rickert 2:38, "If by Chaucer, [the couplet was] cancelled later."
Some further examples
2. The Epilogue of the Man of Law's Tale, or Man of Law's endlink, is extant in 34 MSS, missing from Hg and El--and indeed, from all MSS with the El order. This last fact is important because the endlink names the next speaker in the sequence (at 1179-85: "Seyde the Shipman… / My joly body schal a tale telle" in Riverside). The problem is that the MSS that have the endlink disagree on who speaks. According to most MSS, it's the "Squyre"; 6 or 7 MSS name the "Sompnour"; and only one (unreliable) MS, Se names the "Shipman." Obviously, the speaker that's named corresponds to the tale that comes next in each case, and at least one modern editor (Donaldson) has emended the line to indicate the next tale in the Ellesmere order (hence "Wif of Bath" for "Squyre or "Sumpnour"). Chaucer probably wrote the endlink himself (perhaps on a separate sheet) and later cancelled it (Donaldson, "Ordering" 201-02); because the Tales circulated in various stages of revision, however, the endlink was still available for copying in the fifteenth century and found its way into a little over half the extant MSS of the Tales.
3. In most MSS of the Wife of Bath's Prologue, the Wife of Bath thanks God that she has wedded five husbands at 3.44 and welcomes the sixth at 3.45. In fourteen MSS, however, these two statements are separated by the following lines:
Yblessed be God that I have wedded fyve! 44
Of whiche I have pyked out the beste, 44a
Both of here nether purs and of here cheste.
Diverse scoles maken parfyt clerkes,
And diverse practyk in many sondry werkes
Maketh the werkman parfyt sekirly;
Of fyve husbondes scoleiying am I. 44f
According to Manly and Rickert 2:191-94, these lines are late insertions by Chaucer in WBPro, as are the following:
I bar hym on honde he hadde enchanted me--
My dame taughte me that soutiltee--
And eek I seyde I mette of hym al nyght,
He wolde han slayn me as I lay upright,
And al my bed was ful of verray blood;
'Byt I hope that ye shal do me good,
For blood bitokeneth gold, as me was taught.'
And al was fals; I dremed of it right naught,
But as I folwed al my dames loore,
As wel of this as of othere thynges moore. 575-84
For certes, I am al Venerien
In feelynge, and myn herte is Marcien.
Venus me yave me my lust, my likerousnesse,
And Mars yaf me my sturdy hardynesse; 609-12
Yet have I Martes mark upon my face,
And also in another privee place.
For God so wys be my savacioun,
I ne loved nevere by no discrecioun,
But evere folwede myn appetit,
Al were he short, or long, or blak, or whit;
I took no kep, so that he liked me,
How poore he was, ne eek of what degree. 619-26
[Of Eve's transgression:]
For which that Jhesu Crist hymself was slayn,
That boghte us with his herte blood agayn.
Lo, heere expres of womman may ye fynde
That womman was the los of al mankynde. 717-20
Note that in each case the insertions are self-contained (i.e., if you remove them the poem may be poorer, but it still makes sense).
4. ClT 4.1212a-g, the so-called Host's Stanza, does not appear in all MSS (though it does appear in El and Hg), either because it was a late addition or because Chaucer had cancelled the stanza and decided to reuse the material for the Host's response to the Tale of Melibee (cf. ProMkT 7.1891-94).
5. In some MSS, it is the Host who interrupts the Monk's Tale (ProNPT 7.2767). In others (including El and Hg), however, it is the Knight. Moreover, some MSS (El, for instance, but not Hg) also give the Host and the Knight a pair of speeches on telling happier stories (7.2770-90). What happened? The substitution of Knight for Host, at least, can be explained as follows. You will recall that earlier in the fragment, the Host had interrupted Chaucer's Tale of Sir Thopas with the words, "Namoore of this, for Goddes dignitee…Thy drasty ryming is nat worth a toord!" (Thop 7.919-30). Presumably what happened was this: originally Chaucer had the Host interrupt the Monk; then he decided to have the Host interrupt the Tale of Sir Thopas. Evidently, he considered two interruptions by the Host in one fragment excessive, so he turned the job of interrupting the Monk over to the Host (who was also closer to the Monk in rank, and appears to have known one of the modern instances, Pedro of Cyprus, personally).
6. While we're on the subject of the Monk's Tale, it's worthwhile to note that in some MSS (including El and Hg), the so-called modern instances (MkT 7.2375-462 = the tragedies of Pedro of Spain, Pedro of Cyprus, Bernabò Visconti, and Ugolino) follow the tragedy of Croesus, instead of being sandwiched inbetween the tragedies of Zenobia and Nero (where they're printed in the Riverside). This is a bit odd, because the tragedy of Croesus is actually a very nice place to stop (see especially the generalizations on tragedy at 7.2761-66). Whichever placement was Chaucer's final intention, putting the modern instances at the end of the tale does accomplish two things: (a) it restores the sequence to something approaching chronological order (in which ancient history precedes modern) and (b) it gives the Host (or the Knight) a reason to interrupt the Tale: apparently, the Monk is going to keep on telling tragedies, right up to the present day!
7. The Epilogue to the Nun's Priest's Tale, in which the Host comments on the Priest's virility (e.g., EpNPT 7.3448: "I-blessed be thy breche, and every stoon!"), does not appear in most MSS (including El and Hg), presumably because Chaucer decided to reuse the material for the Monk's Prologue, in which the Host makes a similar point about the Monk (compare, for instance, EpNPT 7.3451 with ProMk 7.1945). As with make the Host's Stanza (4) and the interruptions to Thop and MkT (5), it would appear that Chaucer was striving to avoid monotony.
8. Up to this point, all of our examples have dealt with revisions that editors attribute to Chaucer himself. Scribes, however, also make "improvements" from time to time. The following example (from Benson, "Order" 105-06) is an interesting case, because it illustrates the difficulty of assigning an order to the Tales.
At Ellesmere fol. 123r, the Squire's Tale is interrupted as follows:
In feith Squier / thow hast thee wel yquit
And gentilly / I preise wel thy wit
Quod the Frankeleyn / consideryng thy
yowth SqT 5.673-75
"The last syllable," as Benson notes,
is a regular pentameter ('considerynge' is trisyllabic in Chaucer's verse; cf. Womanly Noblesse, 18). In the Hengwrt MS, however, the last line has only nine syllables:"
Quod the Marchant / considerynge thy yowthe [Hg fol. 137v]
Later, at line 696, the Ellesmere again has a pentameter:
What Frankeleyn / pardee sire wel thou woost
Again the Hengwrt version lacks a syllable:
What Marchant / pardee sire wel thow woost
Three lines later, in line 699, Ellesmere reads:
That knowe I wel sire / quod the Frankeleyn
This time Hengwrt adds a syllable to the Ellesmere pentameter:
That knowe I wel sire / quod the Marchant certeyn
Fifteenth-century versifiers did not always aim at, or achieve, exact pentameters in every line, but it is beyond probability that the original composer would have missed all three times, whereas a later scribe adapting these lines would have hit on a regular pentameter every time. Nor is it clear how he would have happened to hit upon the Franklin, whom this link fits so very well, to replace the Merchant, to whom it has no very clear relation....
There can be no doubt that the Hg versions are adaptations of the versions that appear in El.
The point, though, is not that the Hengwrt scribe was stupid and the Ellesmere scribe was smart--on the contrary, most paleographers agree that both MSS were made by the same man. Rather, the need to fiddle with the identity of the Squire's interrupter probably arose because the Hg scribe did not initially possess the Squire-Franklin link (see the cramping on Hg fol. 137v) and therefore had to guess at which tale should follow the Squire's. He guessed wrong and copied the Merchant's Tale on the next leaf of his booklet. "Unfortunately, when [the] link arrived [it] proved an embarrassment…. The scribe was able to use [it], but only by dint of changing the names of the pilgrims to accommodate the links to the tale order already copied into booklet 4" (Hanna, "Hengwrt and the Canon" 152). Hence the mismetering.
Unless otherwise noted, The Canterbury Tales are quoted from Larry D. Benson (gen. ed.), The Riverside Chaucer, 3rd ed. (Boston: Houghton Mifflin, 1987). For manuscript abbreviations (e.g., El for Ellesmere, Hg for Hengwrt), see Riverside Chaucer 1118-19.
Benson, Larry D. "The Order of The Canterbury Tales." Studies in the Age of Chaucer 3 (1981): 77-120.
Cooper, Helen. "The Order of the Tales in the Ellesmere Manuscript." The Ellesmere Chaucer: Essays in Interpretation. Ed. Martin Stevens and Daniel Woodward. San Marino: Huntington Library and Tokyo: Yushodo, 1995. 245-61.
Donaldson, E. T. "The Ordering of the Canterbury Tales." Medieval Literature and Folklore Studies: Essays in Honor of Francis Lee Utley. Ed. Jerome Mandel and Bruce A. Rosenberg. New Brunswick, nj: Rutgers UP, 1970. 193-204.
Fisher, John H. "Animadversions on the Text of Chaucer, 1988." Speculum 63 (1988): 779-93.
Hanna, Ralph, III. "The Hengwrt Manuscript and the Canon of The Canterbury Tales." Medieval Literature: Texts and Interpretation. Ed. Tim Williams Machan. Medieval and Renaissance Texts and Studies 79. Binghamton, ny: Center for Medieval and Early Renaissance Studies, 1991. 17-39. Rept. Pursuing History: Middle English Manuscripts and Their Texts by Ralph Hanna III. Figurae: Reading Medieval Culture. Stanford: Stanford up, 1996. 140-55.
[M-R.] Manly, John M., and Edith Rickert, eds. The Text of The Canterbury Tales, Studied on the Basis of All Known Manuscripts. 8 vols. Chicago: University of Chicago Press, 1940.
Owen, Charles, A. The Manuscripts of The Canterbury Tales. Chaucer Studies 17. Cambridge, Engl.: Brewer, 1991.
Pearsall, Derek. "Authorial Revision in Some Late-Medieval English Texts." Crux and Controversy in Middle English Textual Criticism. Ed. A. J. Minnis and Charlotte Brewer. Cambridge, Engl.: Brewer, 1992. 39-48.
Ruggiers, Paul G., ed. Editing Chaucer: The Great Tradition. Norman, Okla.: Pilgrim, 1984.