The reference is: Sri Aurobindo Archives and Research, December 1986, Vol. 10, No 2, pp. 169-88. The Table of Corrections, pp. 189-235 is not reproduced.
Was not the 1950-1951 edition of Savitri approved by Sri Aurobindo?
During 1979-86, for eight years, the Archives editors Peter Heehs and Richard Hartz examined all the available drafts of Savitri and, based on certain editorial policies formulated by them, prepared a comprehensive list of changes that should be introduced as new readings in place of the ones present in the earlier printed editions. These readings or corrections are essentially of two types: i) transmission errors arising because of the composition passing through several stages, and through several hands, including the preparation of fair copy, typing, proofreading, and these going back and forth a number of times; ii) emendations of the text and punctuation keeping in view what could possibly represent Sri Aurobindo’s final intentions. It is also said that this proposed critical edition was prepared under the supervision of Nirodbaran and KD Sethna [Amal Kiran], they taking the final decision regarding the recommended changes; further, it had the “approval” of Nolini Kanta Gupta himself. “These three men were associated with Sri Aurobindo in his writing, revision and publication of the poem,” inform us the editors in their introduction. It continues, “the present edition has the endorsement of Nirodbaran and KD Sethna, who have seen and sanctioned all the changes introduced in the text.” The authority or claim by “these three men” is solely ascribed to their direct association with Sri Aurobindo. That seems to be their lone qualification, which rather sounds odd if not funny here.
But then such is the unfortunate justification provided by the editors in support of the departures in Savitri, these insisted by them, such large-scale crucial changes and revisions in the absence of the author. It is also said that the author is not responsible for every word printed in his books! But can anyone actually claim the right or entitlement to give consent for making changes in what he left behind? Procedurally and in principle that becomes perplexing when, in reality, nobody had anywhere given any authority or power for “approval” to anybody, none; it plainly amounts to grabbing or appropriating the right to make changes.
And, again, what are the qualifications of “these three men”? according to the note of the editors, nothing but their “association” with the author of the work. But that itself looks rather strange, if not unusual, association giving authority; at the best it might provide some credibility to what they say. If we have to give a rough though not inappropriate example, it is like an experienced compounder in a dispensary prescribing medicine to patients—only because he had association with a qualified England-educated doctor who is no more living! It is certainly reading in the darkness of the night an inspired and revelatory text, reading in scant light of the clay-lamp of mind instead of the sunlight of the luminous day.
Here it might be interesting to look into the basis of “these three men”, though best among us, setting themselves up to approve what has been set in front of them to approve. However, Nolini Kanta Gupta passed away early in 1984 and it all devolved only on Nirodbaran and Amal Kiran. Nevertheless, any “approval” from whomsoever it might be would, in the strictest sense, carry no content unless each and every entry is examined by him, with all the data in his hand, he going through all the entries and details himself. Simply seeing what is shown is never sufficient. It is expected that “these men” will not merely go by what is presented to them in a meeting when in an hour or so dozens of them are disposed of; this is particularly so for them who were directly involved in this exercise. It should also be pointed out that the examination of this comprehensive archival research by them was not done at any early stage of the present work when it was in progress, done before or while preparing the critical edition of Savitri. It is a post facto examination even as the Table of Corrections was already published, in December 1986; this examination by Nirodbaran and Amal Kiran is thus just a ratification. The exercise already has the tinge of rationalizing the archival work: the Table of Corrections had become a fait accompli. Surely, things cannot be approved on the basis of pre-judgement. More importantly, however, for a composition like Savitri any ‘examination’ has to be in the quiet of the mind; it has to be by deeply identifying oneself with the text, and by invoking the inspiration that brought it down. It is in its light that the understanding should get the guidance. This does not happen in an argumentative discussion or in a debate.
But let us first go back to the very beginning of this whole exercise. When Jayantilal Parekh, the then in-charge of the Archives, spoke in the late 1970s to Nolini Kanta Gupta about the revisions in Savitri and their incorporation in a printed edition which would become authoritative, the latter had simply said: “If Nirod approves.” This was of course much before his passing away in 1984; it was about when the work was proposed to be taken up, in the late 1970s. But everything is contained in it, in that pregnant phrase: “If Nirod approves”. Then the suggestion that the Archival editing had the sanction of Nolini Kanta Gupta becomes somewhat dubious, misleading. Significantly, however, the occult responsibility vis-à-vis the approval was passed on by Nolini Kanta Gupta to Nirodbaran. That begs a question.
If Nolini Kanta Gupta had the final authority among “these three men”, as is purported by the Archives, then what locus standi even for Nirodbaran? none, and none at all for Amal Kiran. According to Nolini Kanta Gupta’s statement,—“If Nirod approves”,—Amal Kiran does not come anywhere in the picture. In fact this whole business of “approval” becomes improper, becomes unauthorized, it also becomes in terms of principles unacceptable. And are not the editors of the aborted critical edition of Savitri mixing up facts in terms of the time sequence? But our concern is to ask the following question: where is the question of Nolini Kanta Gupta “approving” the changes if he had left he matter to Nirodbaran? This whole theory, this idea, of “these three men” and their “association” with the work of Savitri seems to be there only to obfuscate the issue. Association may bring respect and reverence, particularly in a spiritual context, but not necessarily authority, it cannot give adhikāra, and spiritual adhikāra is an altogether different thing. True,
His life, a Virgilian song to the august sun,
A canticle and a prayer brightly enriched
In meaning of the birth of the Supreme.
Yet it does not endow Nolini Kanta Gupta any power to approve or not to approve changes in the writings of his Master, in the least; in fact he would never do that. The point is, this curious theory of association should be dismissed in the context of the Savitri-editing. One may get help from the association, the direct contact, but one cannot go by it.
It should also be mentioned that, when this eight-year work was going on in preparing the critical edition, the proposal was to put certain readings in the main text and their alternatives as footnotes. But in the Revised Edition, 1993, finally presented on the basis of these ‘researches’ and ‘approvals’, we find that there are no footnotes, no alternative readings anywhere in the book. Let us take an example, from Book Four Canto Two, towards the end of the canto: (p. 367 in both 1972 and 1993)
In the 1951 and 1993 editions we have the line
Earth nursed, unconscious still, the inhabiting flame.
But in the 1954 and 1972 editions we have instead
The wide world knew not yet the inhabitant flame.
The proposal was to put this line as a footnote. This has not happened. By the way, this example itself provides reasons to suspect the “approval”-theory, in fact the very methodology of doing things. It is said that Amal Kiran and Nirodbaran approved the changes suggested by the Archives. So the situation is something like this. What was in the 1951 edition that was changed by Amal Kiran in the 1954 and, later, retained in the 1972; but, again, under his own “approval” it was reverted to the 1951 reading. We have to see these fluctuating positions when it is proclaimed that the Mother had “approved” this and the Mother had “approved” that, thus making her own position dependent upon these factors which kept on fluctuating. Can that be so? is that so? We can’t say that the Mother had “approved” such changes which themselves look so uncertain. In fact, the question to be asked is: Is the Mother’s “approval” going to sway at different times, in 1950, 1954, 1972? The answer “Yes” to it will be preposterous. It will be not only in terms of editing Savitri; it will be for Savitri itself, for Sri Aurobindo’s Savitri. And how can the Mother at all approve anything else than what Sri Aurobindo himself had “approved” in 1950, when his work had gone for publication at that time? Part One of Savitri was published in September 1950 before his withdrawal in December that year; Part Two and Part Three came out in May 1951, possibly when the manuscripts had gone to the Press during his own lifetime. Was not that edition of Savitri, the 1950-1951 edition approved by Sri Aurobindo?
But here let us first look into the Savitri-work done by the Archives team. Here is their introduction. ~ RYD
1: Preliminary Remarks
The critical edition of the text of Savitri, announced in these pages in December 1981, is ready for publication. The new edition is the result of eight years of careful checking and rechecking of the original manuscripts. This work was done under the supervision of Nirodbaran and KD Sethna [Amal Kiran], and with the approval of the late Nolini Kanta Gupta. These three men were associated with Sri Aurobindo in his writing, revision, and publication of the poem. The present edition has the endorsement of Nirodbaran and KD Sethna, who have seen and sanctioned all the changes introduced in the text. For the convenience of the readers of Savitri, these changes have been listed in the Table of Corrections that follows this article.
The changes are of two types: (1) correction of “transmission errors”, that is, mistakes introduced by copyists making transcriptions for Sri Aurobindo’s use, which were not caught by him at any later stage, and (2) emendation of words or marks of punctuation that, in the opinion of the editors, do not represent Sri Aurobindo’s final intentions. Both types of change will be examined in detail below. But before entering into the largely technical discussion that this examination will entail, it is necessary to answer two questions that undoubtedly will trouble many admirers of Sri Aurobindo’s epic. First, was an editorial revision of Savitri really necessary? And second, does not any such revision amount to a corruption of a text that the author himself prepared for publication?
The answer to the first question is Yes; to the second one, No. To understand the reasons for these answers, one must first have a clear idea of how Savitri was written and revised. The next section of this article will attempt to give a brief account of this. Secondly, one must know something about the principles of textual critism. The third section of the article will attempt to make this subject accessible to the non-scholarly reader. Both topics have been dealt with in the series of articles entitled On Editing Sri Aurobindo, published in this journal between December 1980 and December 1981 (Vol. 4, No. 1 to Vol. 5, No. 2). In the present article, the treatment will centre around the specific problems presented by the text of Savitri. [1]
2: The Writing and Revision of Savitri
Sri Aurobindo worked on Savitri from August 1916 [2] until the month before his passing in 1950. During these thirty-four years the poem grew from a somewhat extended narrative into an epic of almost 24,000 lines. The genesis of Savitri has been sketched elsewhere, [3] and it will be considered in detail in the bibliographical note to the new edition. The present article is more concerned with the way Sri Aurobindo wrote and revised his epic than with the stages of its development.
Practically all of Sri Aurobindo’s works, Savitri included, were originally handwritten in notebooks or on loose sheets of paper. Sri Aurobindo was a fluent writer. There is little evidence of false starts or backtracking in his manuscripts. After writing a good amount—a few pages of prose or a page or two of verse—he went back and revised. Additions and alterations were handwritten between the lines and in the margins. Then writing and revision would go on together until a good stopping place was reached. At this point a fair copy would be made, generally by hand. This would be revised while the work of composition continued. When the fair copy became dense with corrections, a new one would be made. Perhaps this time the copy would be typed. Each new copy would be revised until a version that Sri Aurobindo considered final was reached. But many drafts marked “final” show signs of revision, sometimes quite extensive revision. The author’s quest for verbal perfection continued through the correction of galley and page proofs. He even made minor revisions in copies of his books that he opened at random.
Savitri began as a narrative poem something on the lines of Love and Death. In its first dozen or so drafts the entire work did not exceed fifty typewritten pages. The early versions were divided into a small number of cantos. These were primary divisions and not subdivisions of books. During the thirties part of the first canto was developed into three “books” consisting each of many cantos. For a number of years Sri Aurobindo devoted what little time he could spare for Savitri to the enlargement of these three books, which form the first “part” of the poem. By 1944 a fair copy of this part had been completed. This draft, handwritten by Sri Aurobindo in two columns on both sides of standard-sized bond paper, was then revised. The first layer of revision was done by Sri Aurobindo in his own hand. Later, as his eyesight began to fail, he took the help of a scribe, [4] who wrote extensive additions and alterations from Sri Aurobindo’s dictation.
After the first part had been revised, the scribe began to copy it by hand into a 393-page ledger. [5] This transcription was then read out to Sri Aurobindo and revised orally. After this, a typed copy incorporating the new revisions was made. This was revised in its turn; sometimes two stages (top and carbon copies) or even three stages of revision exist.
Once the first part had been brought to a satisfactory state of fullness and perfection, Sri Aurobindo turned his attention to the cantos that eventually became Parts Two and Three. These cantos passed through fewer drafts than those of Part One. The present Book 4 and part of Book 5 existed in a copy in Sri Aurobindo’s hand in a notebook which was later used for the scribal copy of much of the rest of the poem. The manuscript for the balance of Book 5, the first cantos of Books 6 and 7, and almost the whole of Books 9 to 12, consisted of loose sheets handwritten by Sri Aurobindo at a relatively early period. When the dictated revision became extensive, the reverse of these sheets was used by the scribe for writing out the revised version of long passages or whole cantos. A fair copy of these cantos was then made by the scribe, which was typed and revised again in the usual way. Book 6, Canto 2, and Book 7, except for the first canto, did not form part of the original scheme of the poem. Lines for Book 6, Canto 2, “The Way of Fate and the Problem of Pain”, were first written by Sri Aurobindo on the chit-pads he was using in the mid 1940s for the revision of Savitri. Much of this canto was composed orally. Among the passages dictated were several long additions to the typed copy. These were the last lines of Savitri he composed. The first draft of Book 7 (from the second canto onwards) is found in a notebook that also contains writings dated 1947. The scribal version was not copied directly from this almost illegible manuscript, but was dictated with reference to it. Revision of the typed copy of this book was very extensive. The present Book 8, “The Book of Death”, which existed only in a very early version, was set aside while work on the other books proceeded. Ultimately it was hardly worked on at all.
On 15 August 1946 extracts from Savitri appeared in KD Sethna’s article “Sri Aurobindo—A New Age of Mystical Poetry”, published in the Sri Aurobindo Circle. This publication had been approved by Sri Aurobindo in March of the same year. Shortly thereafter separate cantos began to be printed, either in journals published from Calcutta and Bombay by groups connected with the Ashram, or in small fascicles printed by the Ashram press. The proofs of these journal-instalments or fascicles were read out to Sri Aurobindo and corrected by him. He also heard and corrected the printed text of each of the cantos that were published. In 1950, before all the cantos of the poem had been released separately, the whole of the first part was printed in book form by the Ashram press. The proofs of this first edition were read to Sri Aurobindo, and he made some changes and additions. No other edition of Savitri was printed during Sri Aurobindo’s lifetime.
3: Principles of Textual Criticism
The new edition of Savitri has been called a critical edition. Readers unfamiliar with bibliographical terminology might find this term confusing. It does not mean an edition prepared in accordance with some theory of literary criticism. This would imply that the editors of Savitri felt free to modify anything in the text that did not agree with their preconceived notions. Nothing could be farther from the method actually used. A critical edition is rather one prepared according to the principles of the discipline known as “textual criticism”. Briefly stated, the task of the textual critic is to prepare an ideal text of a written work from whatever manuscript or printed copies of it are available. The technical term is preferable to “textual editing”, since editing in its ordinary sense often includes the rewriting of material to suit the needs of a publisher—something the textual critic never does. But in this article the high-sounding title “textual critic” has been abandoned in favour of the more familiar “editor”.
Textual criticism is a complex and specialised field, but it is easy to dwell too much on its complexity. The editor of a literary text must apply a technical knowledge, but his greatest asset is enlightened common sense. Behind the fine points of his discipline lies a simply stated and straightforward aim: to transmit the works of an author in error-free texts that represent the author’s final intentions. A great deal hangs on this last phrase. Its essential meaning, however, is so simple as to seem self-evident. The duty of the editor is to present the text exactly as the author would have wanted it presented. The editor’s whole difficulty is to apply this principle to the countless textual situations that arise while preparing a text for publication. And as an authority has written, “every textual situation is unique”, so that even those rules and methods that can be formulated are at best only guidelines that must be applied with discrimination in each separate case.
One rule, however, must be adhered to as closely as possible. A textual editor must base his decisions on objective textual evidence, and not on subjective factors such as aesthetic preference. In other words, he must follow what the author wrote or caused to be written on his manuscripts unless there are overriding reasons for him not to do so. If a reading clearly is defective, he may emend it; but before doing so he should mentally examine every other possible solution, and satisfy himself that his emendation is what the author actually intended to write. And he must give the manuscript reading in a note so that readers can corroborate, or reject, his editorial judgment.
Subjective judgment does have a role to play in textual editing, but it is a supporting role. The editor must have enough critical sense not to print an inferior reading just because “according to the rules” it is the correct one. His literary sensibility will help him confirm that the decisions he has made on objective grounds are sound. And when, as sometimes happens, manuscript evidence is not clear enough to resolve a disputed point one way or the other, the editor will have to use his critical mind in order to arrive at a decision.
4: Some Working Assumptions
Since every text is unique, few general editorial principles can be formulated. But there are constants in each author’s writing style, and a logic behind the composition and revision of each of his texts. These permit the formulation of certain working assumptions that will underlie the editor’s work with that text. Here are some of the working assumptions that governed the editing of Savitri. Most of them will seem self-evident:
- 1. Every legitimate alteration of the text was made by Sri Aurobindo or someone working under his instructions.
- 2. All legitimate revisions, including oral ones, were written down on paper. No revisions were carried out according to Sri Aurobindo’s oral instructions that were not marked on manuscript, typescript, proof, etc. [6]
- 3. Copies made for Sri Aurobindo were supposed to be exact copies. At most copyists were permitted to supply necessary punctuation where it lacked, or to regularise apparently defective punctuation, capitalisation, etc.
- 4. Any non-trivial alteration of a copy not made by Sri Aurobindo may be considered a corruption of the text.
5: How the Editors Worked
The first task of the editors of the critical edition of Savitri was to take a full inventory of the manuscripts of the poem. This formidable task once completed, the various stages of writing and revision had to be arranged in sequence. It is usually easy to determine which of two copies of something written by Sri Aurobindo is later than the other. In passing from one draft to another, he first made a fair copy of the most recently revised version, or else had a disciple make a typed copy for him. This copy would follow its predecessor—closely if it was done by him, exactly (discounting copying errors) if it was done by another. He would then revise the clean copy. His work of revision consisted almost always of alterations and additions; he rarely cancelled anything outright. If he did not like something he had written, he was more likely to expand it than delete it. This means that his later drafts are almost always longer and more developed than his earlier ones. To determine the general sequence of a group of drafts, all the editors had to do was place the shorter, less developed versions towards the beginning, and the longer, more elaborate ones towards the end. The exact sequence could be found by matching the revised version of one copy with the unrevised version of the copy that followed it. (This unrevised version could be read by ignoring for the moment the corrections and additions written in ink above the original transcript.)
Once the sequence of manuscripts had been established, the editors had to decide which version of the poem would be used as the basis of the critical edition. This is a very important decision in any work of textual criticism, since the version selected—known as the “copy-text”—inevitably has a shaping influence on the final form of the critical text. When the work being edited is one that appeared in print during the author’s lifetime, it is common practice to choose either the first edition or the last edition revised by the author. For two reasons this approach could not be adopted by the editors of Savitri. First, the poem was not completely published during Sri Aurobindo’s lifetime. Parts Two and Three were seen through the press by disciples. Secondly, the editors realised at an early stage of their work that their chief task would be to find and correct errors resulting from copying mistakes made by persons other than the author. Most such “transmission errors”, as they are termed, [7] would have occurred while the poem was still in manuscript. To be sure of catching all of them, the editors had to go back to the last stage that Sri Aurobindo alone was responsible for. In cantos composed by hand, this was his final handwritten draft; for the few cantos that were composed orally, it was the transcription of the original dictation. The manuscripts representing this stage, which together are referred to somewhat loosely as the “final version” of the poem, form the copy-text of the critical edition of Savitri. The editors referred to earlier handwritten versions only when a reading in the last handwritten version was illegible or doubtful. Later versions were, however, referred to constantly, since they contain innumerable authorial revisions and additions, most of which were written by the scribe from Sri Aurobindo’s dictation.
The editors called the earliest manuscript that they read in each canto Version A. Subsequent stages were referred to as Version B, Version C, etc. The number of version differs greatly from canto to canto. Book 8 has only one version. The cantos of Part One (Books 1-3) have as many as seven:
A—Handwritten and hand-revised double-column draft.
B—Lengthier additions to (A) handwritten by Sri Aurobindo or his scribe on chit-pad sheets or in separate notebooks.
C—Scribal transcription of (A) + (B) handwritten in a ledger and revised by dictation.
D—Typed copy (or typed copies, designated D1, D2 . . .) of (C).
E—Proofs of (F).
F—Fascicles or journal-instalments consisting of one or more cantos, revised by dictation.
G—Proofs of the first edition.
Once the sequence was established, the editors proceeded to the next and most laborious step—collation. Collation means to read the different versions of a work against one another in order to locate any differences between them. The ideal means of doing this is to duplicate the process of transcription, reading each stage of the poem against the one that preceded it. Unfortunately, it would have taken far too long to read all the versions of such a long poem as Savitri in this way. Limited both in time and in manpower, the editors were obliged to look for a shortcut. The method they settled on, which itself took them some seven years to carry out, was to read the first stage (Version A) of each canto against the text of the last printed edition, i.e. the Centenary Edition of 1972. When any difference between these two versions was noted, the source of the change or addition was sought among the stages that intervened. If the transcription of the text was carried out correctly all the way through, no irregularities would be noticed. if, however, a mistake of copying, typing, composing or editing took place anywhere along the way, the discrepancy would be visible. Once an error was identified and its source discovered, it could be dealt with according to pre-established guidelines.
This shortcut method was obviously not so foolproof as a complete collation would have been. For example, deliberate authorial changes made after Stage A that were not carried over to subsequent stages would not be caught unless every stage were read. Fortunately, this sort of copying error was not common, since all deliberate changes were marked clearly on the manuscript. The editors considered it sufficient to make a spot check for missed revisions. If any slipped through, it is unlikely that they had any significant effect on the text.
6: What is a Transmission Error?
The reader will have noticed that the editors spent much of their time looking for sinister things known as “transmission errors”. Just what does this term mean? Very briefly, a transmission error is one made by someone other than the author during the copying or printing, that is the “transmission”, of a text. The author evidently is not responsible for such mistakes. It should be noted that changes made by the author while making copies of a work for his own use are not considered transmission errors. Authorial copies are in a sense part of the transmission of the work, but they differ in one important respect from copies made by another. An author can and often does make deliberate changes while recopying. [8] Therefore an authorial copy cannot be expected to agree with what precedes it in every particular. But a copy made by another, whether handwritten, typed or printed, is supposed to be an exact duplicate of what preceded it. This makes the finding of transmission errors a relatively easy task. If the previous stage has been preserved (as were almost all of the innumerable stages of Savitri), the correctness of the copy can be checked. Unauthorised differences between one stage and the next can only be due to a mistake of copying, typing or printing.
We have seen that the text of Savitri went through three stages of transmission in which Sri Aurobindo had no hand:
- 1. The copying of the final handwritten version of the poem by the scribe.
- 2. The typing of this handwritten version.
- 3. The typesetting of this typed version.
A copying mistake made during any of these stages would count as a transmission error. If such errors were not noticed by the author, they became parts of the text. Careful comparison of successive stages of the manuscripts of Savitri has shown that the human instruments Sri Aurobindo used for the work of copying and typing his poem did indeed make mistakes. There is nothing unexpected or reprehensible in this. If one thinks of the hundreds of pages of manuscript, written in a difficult, sometimes barely readable hand; of the number of versions of the different cantos, some of which were being worked on concurrently; of the rapidity at which the work was done, both to keep pace with Sri Aurobindo, and to satisfy his anxiously waiting readers, one realises that the errors were very few. Still, they number in the hundreds, and each of them is a misrepresentation of what Sri Aurobindo wrote. All of them had to be removed from a text that claims to represent Sri Aurobindo’s final intentions.
Here an important consideration intervenes. Sri Aurobindo either went through or had read to him most of the copies that contain the so-called transmission errors. How can he not have noticed them? In fact he did notice and correct many errors. Sometimes, remembering what the original reading was, he altered the text back to what it had been. Sometimes he changed the offending word or words to something entirely new. Often enough, however, he does not seem to have noticed the error at all. Is it possible for an author, in particular an author who lived and wrote from an elevated state of consciousness, to miss errors in something he himself had written? Sri Aurobindo, at least, did not think it impossible. Once when his secretary pointed out an obvious transcription error that had not been corrected on a proof Sri Aurobindo had gone through, the latter replied without the least embarrassment: “I did not notice this mistake while correcting.” No one familiar with the complicated process of writing, revising and correcting proofs will be surprised to learn that even Sri Aurobindo allowed errors of transmission to pass. After a certain stage in the preparation of a book, many authors are more concerned about the general shape of the chapter or canto, about the overall movement and flow, than about details of phrasing and punctuation. Sri Aurobindo certainly worked in this way. He once wrote to his secretary: “I have corrected [the pages sent] from the point of view of substance and language only—except where a typographical mistake met my eye.” If an author who has brought his work to an advanced stage of revision looks at details at all, it is generally only in passages that were recently written or rewritten. Familiar passages will of course be given a cursory glance, but unless the error is striking—something that makes the sentence ungrammatical or ruins the metre of a line—it is possible for it to slip by.
But still, if an author passes a certain version, is it not the duty of his editors to print it as it is, error or no error? No editor wishing to present a text as the author intended it to be presented could afford to do so. Transmission errors are by definition introduced by someone other than the author. To accept them would mean permitting another person to have a share in the writing of the work. Moreover, most transmission errors are manifestly inferior to what the author originally wrote. Why accept a reading that is both corrupt and bad when a genuine and good one is available?
All this seems obvious enough. But when the corrupt reading has gained currency by being in circulation for a long time, the matter becomes more difficult. An editor working on an important text years after its author, and everyone who had assisted him in its preparation, had passed away, would find the question of transmission errors a vexing one indeed. Fortunately, the critical edition ofSavitri has been prepared with the approval of the disciples who assisted Sri Aurobindo in its preparation, and who, after Sri Aurobindo’s passing, have made all necessary editorial decisions in connection with the text. It was they who, in the course of this exacting work, committed most of the errors listed in the Table of Corrections. These men have agreed that all the transmission errors that have crept into the text of Savitri have to be corrected.
6: Examples of Simple Transmission Errors
Most transmission errors in Savitri were straightforward copying mistakes. An example occurs in Book Two, Canto 7, line 507 [p. 219][9]:
Breathed in a laxed content its natural air. [10]
In Sri Aurobindo’s manuscript (A) and the scribal copy (C) the word italicised here is “lax”. It was typed “laxed” in version D1, and so it has remained until the present. No doubt when the line was read out to Sri Aurobindo he did not catch the superfluous verbal ending, which would not strike the ear when pronounced after the “x” of “lax”. “Laxed”, if accepted, would have to be considered a neologism. While not altogether impossible in the context, it is certainly inferior to the simple and natural “lax”, which is what Sri Aurobindo actually wrote.
A somewhat similar example occurs in Book Seven, Canto 5, line 23 [p. 522]:
Humanity seemed now too proud a state.
In both Sri Aurobindo’s handwritten manuscript [11] and the scribal version, the first word is “Humility”. The typist turned this into “Humanity”, and the error was never caught by Sri Aurobindo. As in the last example, the corrupt reading makes sense, but lacks point. The line is made memorable by the gentle oxymoron of the correct reading.
Sometimes a copying error has resulted in the loss of a whole line. In the handwritten manuscript and scribal copy of Book Seven, Canto 4 [p. 506] the following lines occur:
What profit have I of my animal birth
What profit have I of my human soul?
When the canto was typed the first line was omitted. The repetition of the first six words of the first line in the second evidently caused the typist to miss it out. Sri Aurobindo did not notice the omission; but he never deliberately rejected th9line. The editors of the critical edition have restored it to the text. Twenty-nine other lines in various cantos, which were lost in much the same way, have also been reinstated.
In all the above examples the transmission error was committed by the typist. Many similar slips were made by the scribe. See, for instance, Book Two, Canto 4, line 550 [p. 147]:
All was unrecognised stress and seekings vague.
In Sri Aurobindo’s manuscript, the word modifying “stress” is clearly “unorganised”. It was copied “unrecognised”, and the error remained through several stages of revision. Here again the erroneous reading is grammatically possible, but inferior in sense to what Sri Aurobindo wrote.
Most transmission errors occurred when the passage concerned was being either hand-copied or typed. A few, however, came in at a later stage. In every printed edition (including the 1948 fascicle), line 217 of Book Two, Canto 10 [p. 244] reads:
And slowly wake beneath the glows of life.
The line was added to the poem during oral revision of a typescript. In the transcript of the original dictation, the word italicised here was “blows”. The line was retyped correctly, but misprinted in the fascicle. The fact that so obvious a mistake slipped by Sri Aurobindo twice (while correcting the proofs of the fascicle and of the first edition) shows that he was much more likely to miss transmission errors in passages read out to him than in passages he corrected himself.
So far all the illustrations have been examples of simple and understandable misreadings by scribe, typist or printer. A few of the errors in Savitri, however, surprise one by their unlikelihood. Two such occur in close proximity in Book Three, Canto 3. The manuscript involved has been reproduced as Plate 2. The reader is invited to confirm what the editors were forced to conclude.
Line 245 [p. 323] of the canto was written by the scribe from Sri Aurobindo’s dictation:
As if prolonging in a ceaseless count. [12]
In the next stage the second-to-last word was written “celestial”. Can the scribe have miscopied something he himself had written? No reason for amazement. All of us make such mistakes. And if we look at the poetic quality of the original line and the miscopied one, we get adequate confirmation that “celestial count” is simply a transmission error.
Line 253 [p. 324] presents us with an even more unusual situation. In Sri Aurobindo’s last handwritten manuscript, the line reads:
Its calm immutable stability. [13]
This was copied:
Its luminous immutable stability.
Surely, one might say, this is not just an ordinary transmission error. No one could have read “luminous” in place of “calm”. Evidently it was Sri Aurobindo who wrote “luminous”, or who told the scribe to write it. But there is no stage of revision between the handwritten manuscript and the scribal copy. And we have taken as one of our working premises that every oral revision of Savitri was recorded in ink. “Luminous” is apparently a transmission error. But how could the scribe have committed such a fantastic blunder? The editors, with years of transcribing experience behind them, know that such errors can and do occur. Here the misreading was the outcome of a process having several stages. After correctly reading the word “Its”, the scribe began to struggle with the words that follow. Besides being rather illegible, they are written together without a break. Where does one word end and the next one begin? Staring at such a puzzle is like gazing into the featureless desert. Finally the word “luminous” appears, stretching from the “I” of “calm” into the rolling dunes of the loops that follow. But unfortunately the reading is a mirage. It disappears immediately, allowing the two concluding words to be correctly deciphered. The fact that “immutable” has already lent its initial letters to the mirage is not noticed. The resulting line is, as usual, not impossible, but its questionable metre (a watery fourth paeon replacing a solid iamb in the second foot) betrays it as a transmission error.
These examples have not been chosen in order to hold the disciples who copied Sri Aurobindo’s manuscripts up to ridicule. No one knows better than the editors of the present edition that these men, specially selected by Sri Aurobindo for this difficult and exacting work, did the job better than anyone else could have done. The reader has only to look at the manuscript page containing the two errors just given to realise how taxing it must have been to transcribe it. And this is not an unusually difficult page. The point of the last two examples is that even when the manuscript evidence, by its very improbability, seemed to require for its solution some sort of deus ex machina, such as a lost manuscript or a forgotten verbal instruction, the editors were nevertheless obliged to adhere to the actual physical evidence.
All the examples so far dealt with are “simple” situations because the transmission error remained unchanged once it was committed. Often, however, a complex situation developed during the process of revision. Such situations, for which special treatment is required, are discussed in sections 7 and 8. A few errors fall somewhere between the classifications of “simple” and “complex”. In Book One, Canto 5, line 474 [p. 871, Sri Aurobindo wrote:
Her sealed hermetic wisdom forced from her.
In the handwritten copy the third word was written "hieretic". (It actually seems to have been written "heretic", then changed to "hieretic".) "Hieretic" is not an English word, so it was altered to "hieratic". Sri Aurobindo probably had nothing to do with this change (the routine correction of a misspelling), nor can one be sure that he was involved in the presumed change of "heretic" to "hieretic". In any case what he wrote is "hermetic': (an earlier manuscript confirms the reading of the final version), and this word—better in substance and metre than the erroneous one—has been reinstated in the critical edition.
7: Complex Transmission Errors (Variants)
Simple errors are clear, self-contained and usually easy to set right. Once discovered, the corrupt reading can simply be removed and the correct one put in its place. The context is not disturbed by this localized operation. Unfortunately, not all transmission errors admit of such an easy solution. Sometimes a copying mistake results in a pair of readings, each of which has a claim to be used in the text. In such cases, the difficulty often spills over into the context, further complicating the matter. Such situations have been called by the editors "complex transmission errors". Most result from a transmission error that was caught by Sri Aurobindo, but which was not corrected by him in accordance with the original reading. An imaginary example will make this clear. If an author wrote "The sky is blue-, and a copyist transcribed this sentence "The sky is black", that would be a transmission error. If the author went over the copy and changed "black" to "blue", the error would cease to be. If, however, he changed "black" to "azure", the result would be a pair of readings, "blue" and "azure", both of which originated from the author's mind, and neither of which was rejected by him.
The presence of such a pair of "good" readings forces a difficult choice: which of the two is correct? So stated, the problem can have no final solution. "Correctness" is the sort of thing that people can never agree on. To restate the problem as "which is better" would make it even more insolubly subjective. The textual editor, working from the evidence contained in the manuscripts, and bringing in subjective evaluation only as a subordinate helper, must state the problem in this way: which of the two has a better claim to be used in the text, and which of them should be relegated to the notes? For neither reading can be discarded; both must be treated as "variants". [14]
Which of the two variants should be used? Cannot one formulate a rule that would keep guesswork and subjectivity out of the picture? At first sight it would seem that one could. Since the editor is sworn to respect the author’s final intentions, it would follow that the second reading should in every case be considered superior to the first. Written later, it must necessarily be closer to the author’s last wishes. But in order to be sure that the author preferred the second reading, one would have to assume either that he remembered the first while correcting, or that, if he did not, he refreshed his memory by referring to the previous manuscript on the spot. This assumption is not acceptable. There is no reason to suppose that any author, even one attuned to “overhead” levels of inspiration, can remember each word he ever wrote. And every writer knows that it is often more convenient to rewrite a sentence, or even an entire page, than to stop work and try to find a lost version among heaps of old manuscripts.
There is another important consideration. Very often the second of two versions resulting from a transmission error seems clearly to have less literary value than the first. While an editor must refrain from making purely subjective judgments, his work, if it is not to be a mechanical following of rules, sometimes with ludicrous results, must be supported by a sense of verbal clarity and metrical order, and a feeling for such values as beauty and power of language. The editor must of course try to view these criteria as the author himself viewed them, and not to impose his own standards on the text. If he notices that one version seems preferable to another for rhetorical or aesthetic reasons, he must take this into account. Such subjective evaluations were especially important when the editors dealt with “variants”, as the following examples will show.
8: Examples of Complex Transmission Errors
Complex errors begin as simple ones. In Book Six, Canto One, line 230 [p.421], Sri Aurobindo wrote:
For it has read and broken the wizard seals.
The second to last word was copied “inner”, a word that occurs in the previous line. When the passage was read out to Sri Aurobindo, he noticed the obvious repetition. He therefore changed “inner” to “hidden”, and so it remained throughout the rest of the revision. But the original adjective, “wizard”, is certainly more striking and resonant than “hidden”. That word is acceptable, but certainly if Sri Aurobindo had been asked to choose between “hidden” and “wizard”, he would have selected the word he originally wrote. The editors accordingly have used “wizard” in the text, and placed “hidden” in the notes. “Inner”, which was cancelled by Sri Aurobindo, was of course discarded entirely.
A more complex situation resulted from the miscopying of line 1005 of Book Two, Canto 6 [p. 201]. The line as originally dictated reads as follows:
Birth, death are a ceaseless iteration’s points.
It was typed:
Birth, death are a ceaseless vibration’s points.
Noticing the flatness of substance and rhythm when revising the typed copy, Sri Aurobindo altered the line to read:
Birth, death appear as its vibrating points.
As revised the line reads all right. “Appear as its vibrating points” is certainly preferably to “are a ceaseless vibration’s points”. But in altering the line was not Sri Aurobindo just making the best of a bad job? The whole idea of vibration was brought in by a typist’s error. It never formed part of Sri Aurobindo’s poetic vision. On the other hand, “a ceaseless iteration’s points” is both original and effective. “Iteration” occurs nowhere else in Savitri, it reinforces the sense of “recurrent decimal” in the preceding line, and, in addition, it gives the line a staccato rhythm that supports the intended meaning. Taking all this into consideration, the editors have selected the first version of the line for the text, and placed the last version in the notes as a variant.
It will be seen from these two examples that there is often good reason to prefer an earlier version to one written after a copying error has corrupted the text. The “firm rule” that at first sight seemed so promising—that the hnal version of a line to be written should automatically be used in the text—has proved to be unreliable. But to swing to the other side and say that an earlier version should always be used would also be unwise. For sometimes one cannot avoid using the later version. The examples that follow demonstrate why this is so.
Line 55 of Book Two, Canto 7 [p. 203] was originally written:
Became themselves a danger and a harm.
In the scribal copy. “harm” was written “charm”. When revising, Sri Aurobindo let “charm” pass, but, sensing that something was wrong with the line, he reinforced the sense of the word he had not written by adding a line:
A perverse sweetness, heaven-born malefice.
The new line takes up both the root meaning and the modern implications of “charm”. The original sense, “spell, incantation”, is underscored by “malefice”; the modern meaning, “attractive quality or power”, is taken up by “perverse sweetness”. If “harm” were reinstated in the first line, the second would lose much of its point. But it is too good a line to be dropped. “Charm” was not part of Sri Aurobindo’s original inspiration, but by taking the word up in the revised version of the passage, he has made it part of his poem. The second version of the line has therefore been used in the text, and the original line placed in the notes. [15]
Subjective evaluation has a place in solving complex problems caused by transmission errors; but its place must always be subordinate to assessment of textual evidence. Book Two, Canto 14, line 253 [p. 296] was originally written:
And drunk as with a gold spiritual wine.
It is a fine line; but unfortunately when it was copied “gold” was written “golden”. Even so insignificant a mistake spoiled the delicate metrical balance of the line. Noticing this, Sri Aurobindo altered it to read:
Drunk with a deep golden spiritual wine.
This version of’ the line also is fine; but it is a moot point whether it is “better” than the first. Rather than debate the aesthetic merits of the two versions, the editors determined which one to use in the text through an analysis of Sri Aurobindo’s revision. Faced with the necessity of correcting the line, he could easily have deleted the “en” of “golden”, thus restoring the line to its original form. It apparently did not occur to him to do this. Instead he cancelled “And” and “as” and added the adjective “deep”. This fairly extensive revision represents a good deal of deliberation on Sri Aurobindo’s part. It would have been presumptuous of an editor to discard it simply because he preferred the earlier version. But by printing the earlier version in a note, he preserves a line that did represent the working of Sri Aurobindo’s inspiration when it was written.
9: Emendations
A reading must be emended when it does not represent the author’s final intentions. This rule is easy enough to state, much more difficult to apply. But before going into the intricacies of the solution, we must take a closer look at the problem. When does a phrase, word or mark of punctuation not represent the author’s final intentions? For the author makes these intentions known precisely by what he sets down on paper. To alter something he has written would seem to be an act of second-guessing that could only corrupt the text. What purpose can be served by removing accidental errors of transmission if the editors are allowed to introduce deliberate ones of their own? Certainly emendation should not be resorted to hastily. But sometimes what the author has written is evidently not what he intended to say. All writers make slips. Sri Aurobindo never considered himself an exception to this rule; indeed once or twice he expressed surprise that he did not make more slips than he did. [16]
Happily, those manuscripts of the final version of Savitri that were handwritten by Sri Aurobindo contain very few textual situations that seem to require emendation, and practically all of these involve details of spelling, punctuation, capitalisation etc. (termed “accidentals”) and not words (“substantives”). A representative selection of these, including all of the substantive emendations, is presented below.
Sri Aurobindo made few spelling errors. In the entire handwritten portion of the “final version” of Savitri, there is only one: “alters” spelled as “altars” (Book Four, Canto 4, line 90 [p. 379]): This slip was corrected editorially before the first edition was printed. The spelling of Sanskrit words presented more difficulty, since Sri Aurobindo changed his ideas about transliteration over the years. In early manuscripts he spelled the names of the main characters of his epic “Savithri” and “Suthyavan”. Later manuscripts contain the familiar spellings to which he gave final preference. In copying from earlier manuscripts, the scribe was apparently Instructed to substitute “Savitri” and “Satyavan” for the older spellings wherever they occurred. It may be observed that in the preferred spellings of these two names, the dental “t” is represented by “t” not not “th”, the short “a” is represented by “a” and not “u”, and the circumflex over long vowels has been dropped. These habits of transliteration have been used by the editors of the critical edition to normalise the spelling of three other names that occur only in lines written in earlier manuscripts: “Gundhamadan”, “Alacanunda” and “Dyumathsena”. These have been changed to “Gandhamadan”, “Alacananda” and “Dyumatsena”. The name of Savitri’s father, originally spelled “Uswapathy”, was later written by Sri Aurobindo in two ways: “Aswapathy” and “Aswapaty”. The latter spelling conforms to Sri Aurobindo’s final preference in regard to the dental “t”, and besides occurs in a late manuscript. The editors have accordingly given preference to “Aswapaty”, and made all occurrences of the name accord with this spelling.
Sri Aurobindo’s handwritten manuscripts are carefully punctuated—to the degree that even small irregularities stand out. Two such occur in Book Two: in line 238 of Canto 8 [p. 226, line 30] and line 33 of Canto 10 [p. 239, line 2]. Both lines clearly read better with final commas. In both cases the commas were supplied by the typist, who was evidently authorised to add necessary punctuation as he typed from the scribal copy of the manuscript. There are a few other examples of this sort of emendation, which do not differ greatly from these two.
More complex problems of punctuation resulted from the process of textual transmission. The editors have made two or three emendations in such situations. One may be considered representative. Line 235 of Book Two, Canto 3 [p. 324], “Infallible, leaping from eternity”, was inserted by Sri Aurobindo after a copying error altered the following line. The editors have restored the original form of that line (“Inspired the moment’s thought, the passing act”), but in doing so, they have had to make the line that precedes the one that was inserted end in a comma and not a full stop.
None of the above emendations have any significant effect on the subtance of the text. In all of Savitri, only five substantive emendations have been made in lines handwritten by Sri Aurobindo. All will be discussed here. In Book One, Canto 2, line 281 [p. 18], Sri Aurobindo’s manuscript reads:
In the dire court who life must pay for joy.
Something obviously is wrong here. Sri Aurobindo originally wrote “where life must pay …”, then altered “where” to “who”. Why? He certainly did not intend to make the line unintelligible. The editors, after examining the manuscript, have concluded that “who” was intended to replace “that” in the preceding line. Sri Aurobindo apparently cancelled the wrong word. At any rate the copyist of the manuscript wrote “where” and not “who”. This was evidently the correct way to deal with the situation.
A transmission error produced an unusual situation in Book Six, Canto 1, lines 706-07 [p. 434]. The first line was originally written:
Of work and thought and measured grave delights.
Later a passage was added, the first line of which ended with “height”. When the lines were transcribed, “delights- was copied “delight”. This produced an unwanted rhyme. When revising the canto, Sri Aurobindo got rid of the rhyme by altering line 707 so that the last word became “heights”. It would go against his intention to restore the original reading “delights” in the first line, since that would introduce a rhyme almost the same as the one he had removed. “Delight” must be accepted—an “emendation” inadvertently introduced by the scribe.
Another complex situation that required substantive emendation re-suited from a transmision error made not by the scribe, but by Sri Aurobindo himself. It has already been mentioned that the rectification of such authorial slips must be treated as emendations. Lines 572-73 of Book Two, Canto 7 [p. 218] were originally written by Sri Aurobindo:
A formless void oppressed his struggling brain,
A darkness grim and cold benumbed his flesh.
When copying these lines into the “final” manuscript, Sri Aurobindo wrote
A formless void oppressed his struggling brain,
A darkness grim and cold oppressed his flesh.
“Oppressed” has evidently crept into the second line from the one above it. This is a common sort of copying error, and the repetition stood out clearly. While revising the passage, Sri Aurobindo rectified the situation by altering the first occurrence of “oppressed”, making the lines read:
A formless void suppressed his struggling brain,
A darkness grim and cold oppressed his flesh.
The revision removed an obvious slip; but the original version of the lines seems preferable to the last. As with other “complex transmission errors” the editors have relied on subjective factors to help them decide which lines should be used in the text. After considering the situation thoroughly, they havc kept the original version and relegated the third version to a note.
The last emendation of a handwritten line was necessitated by what the editors consider to be a slip made by the author while revising. All hand‑written versions, except the last, of line 491 [p. 347] of Book Three, Canto 4, run as follows:
And in the pauses of the building brain.
When he copied this line into the “final version”, Sri Aurobindo wrote “twixt” instead of “in”. This word, although somewhat archaic, is perfectly legitimate, and in fact of fairly frequent occurrence in Savitri. But here it does not make sense. The “pauses” of the brain are what come between, or twixt, its ordinary activities. Sri Aurobindo’s intention surely was that it is in these pauses that, as the sequel says, the “thoughts” from “hidden shores” come in and touch the seeker. Perhaps he meant to alter “pauses” when he substituted “twixt” for “in”. At any rate, the unrevised version of the line, as given above, seems to represent Sri Aurobindo’s intentions better than the revised one, and it has therefore been restored to the text.
Words requiring emendation are far more frequent in passages dictated by Sri Aurobindo to his scribe than in passages that he wrote by hand. This is to be expected; it is easy to mishear a spoken word. Often the suspicious reading is a homophone or near-homophone of the intended word. An example is “wants” in Book Two, Canto 10, line 644 [p. 256]:
Master of Nature Who wants her bondslave worked.
Evidently the word Sri Aurobindo dictated was “once”. “Wants” makes sense only if “bondslave” is taken as the object of a transitive “worked”—a highly unlikely possibility.
Emendations have been introduced in every edition of Savitri since the first. Thus far none of them has been listed or otherwise indicated. In the critical edition all significant emendations—both those introduced by previous editors and those, fewer in number, that have been newly introduced—will be clearly indicated by footnotes. The manuscript reading will be given in the note.
10: Punctuation
Most of Savitri was carefully punctuated; but parts of it, especially some dictated portions, were not. If the text of the new edition was to be consistent in matters of detail, as Sri Aurobindo certainly wanted it to be, a certain amount of punctuation had to be added or emended by the editors. Such emendation has in fact been done since the first edition; indeed even the typist and, to a much lesser degree, the scribe added and regularised punctuation. No doubt there was an understanding between them and Sri Aurobindo in regard to this. The editors of all printed editions of the poem also felt called upon to emend punctuation, and much of what they did was undoubtedly necessary. Portions of the poem that were composed and revised orally often lack the most essential pointing, including even full stops. Not that Sri Aurobindo gave no attention to punctuation while dictating. But often he was too preoccupied with getting the words of his inspiration down on paper to be troubled overmuch about the punctuation.
The editors’ general rule in regard to punctuation was not to add or change a point unless it seemed absolutely necessary to do so. Sri Aurobindo’s pointing was relatively light by modern standards. For example, he often did not put commas before or after present participial phrases. The editors have tried to follow Sri Aurobindo’s own practice. In doing so they have removed much previous editorial punctuation that seemed to them unnecessary.
11: Normalisation
Where necessary punctuation was lacking, the editors have supplied it. Where it was noticeably irregular, they have normalised it. Some apparent irregularity is due to the difference in date of the manuscripts making up the “final version”. Some of these were written in the 1940s, others in the 1930s or even earlier. Sri Aurobindo’s practice in pointing changed over the years. He did not begin to use, or approve the use of, the colon to introduce direct speech until relatively late. In most handwritten manuscripts direct speech is introduced by the semicolon. Since the colon was used in printed texts seen by Sri Aurobindo, the editors have substituted a colon wherever a semicolon precedes direct speech in the manuscript.
Other details that have been normalised are capitalisation and spelling. Here the editors’ practice was to let an irregularity pass if it did not stand out. If it did, Sri Aurobindo’s preference was ascertained. By using the Savitri Concordance it was possible to check every occurrence of a word, and to make each occurrence conform to the spelling that Sri Aurobindo preferred. Anomalous spellings were often introduced by the scribe. For instance, he wrote “jinn”, while Sri Aurobindo always spelled the word “djinn”. Both forms are acceptable, but the editors have normalised the spelling to accord with Sri Aurobindo’s own preference.
Capitalisation proved somewhat more difficult to deal with. As with his punctuation, Sri Aurobindo’s habits of capitalisation changed during the years he worked on Savitri. Moreover, when he dictated to his scribe, he seems to have paid less attention to this detail than to any other. For example, in the period when he wrote most of his final manuscript of the poem, he regularly capitalised “Nature” in the sense of cosmic and not individual prakriti. The scribe generally followed this practice, but not with equal consistency. Sri Aurobindo himself had not been so consistent at an earlier stage. Among the manuscripts from an early period which form part of the “final manuscript”, there is one occurrence of “nature” which does not agree with his later practice. This occurrence, and most of those for which the scribe is responsible, were capitalised by previous editors, if not by the typist. The normalisation of this word’s capitalisation has been made complete in the new edition. However, other words often capitalised by Sri Aurobindo have not been normalised, since he himself never arrived at a complete uniformity. In such cases, the editors have generally considered it wisest to respect the variations of the handwritten manuscript. In the case of dictated lines, however, most of the regularising of capitalisation done by the typist or previous editors has been allowed to stand.
[1] Portions of the earlier articles have been incorporated in the present essay.
[2] Sec Archives and Research, vol. 5. no. 2. p. 190. note. Newly found evidence supports the contention of the writer of that article that Sri Aurobindo did not begin to write Savitri in Baroda around 1900. The matter will be discussed at length in the bibliographical note to the critical edition.
[3] See Nirodbaran. Twelve Years with Sri Aurobindo (1972), pp. 162-90; Archives and Research, vol. 5, no. 2 (December 1981), pp. 190-91.
[4] The word “scribe” will be used throughout this article to refer to the disciple who served as Sri Aurobindo’s amanuensis. That Latin term is precise, but unfamiliar. The ordinary English word “secretary” is objected to by many because of its official connotations.
[5] It may be that the scribe began to make the copy before the revision had been completed. Chronological precision is not possible since relatively few of the manuscripts were dated. Here, and in other matters of sequence, only a broad measure of accuracy is possible.
[6] The disciple who acted as Sri Aurobindo’s scribe has explicitly confirmed that this assumption is correct.
[7] “Copying error” would be a handier term; but the errors in question were made while copying by hand, while typing, and while composing for print. All these operations are forms of copying, but that word suggests chiefly the making of a handwritten transcript. The connotations of the technical term “transmission error” are not so limited. The term is defined in the next section.
[8] An author may make a slip while recopying. but if an editor chooses to correct such a slip, he must treat it a.s an emendation. The subject of emendations is discussed in Section 10.
[9] Here and in the examples that follow, line numbers have been counted from the beginning of the canto in the new edition. (The edition will be numbered in this way when it is printed.) The page numbers within square brackets refer to the Centenary Edition.
[10] Words italicised in examples are errors that have been corrected in the critical edition.
[11] See Plate 3, last line.
[12] See the block of lines written sideways in the right margin of Plate 2.
[13] The fifteenth line in Sri Aurobindo’s hand in Plate 2.
[14] In general editorial practice, the term ••textual variant” or “variant” is used to indicate a variation resulting from the author’s deliberate revision of his work. The editors of Savitri have not concerned themselves with such variants. To list every change in the manuscript would be virtually impossible, and there were no printed variants, since Sri Aurobindo saw only part of one edition. The editors have used the term “variant” to indicate each of a pair of readinp resulting from a complex transmission error. The word “alternative”, which might have avoided ambiguity, has been used by them in another sense: pairs of manuscript readings, neither of which was finally selected by the author.
[15] It is unfortunate that the entire textual situation cannot be explained in the notes. It would however be impossible to write an explainion of every situation resulting from a complex transmission error. In this example, without a paragraph such as has been written here, the reader will not know that the variant ending with “harm” should be read without the line that comes next in the text.
[16] In 1936 Sri Aurobindo wrote to a disciple who had discovered a slip of the pen in one of his letters: “Do you mean to say this is the first you have met? I used to make ten per page formerly in the haste of my writing. Evidently I am arriving towards a supramentakaccuracy — spontaneous and careless in spite of the lightning speed of my epistolary movement.” [Letterio Nirodbaran, 14 March 19361 On another occasion he informed a different disciple that he had "no time for accuracy" when writing letters, since they took too much of his time as it was. He advised him to -supply the gaps left by pen-slips for yourself". [Letter to Amal Kiran, February 1931]
