"In order to organize the Government of the Indian Territory. . .": Comparing variants of the 1870 Okmulgee Constitution

The creation of the Okmulgee Constitution is a little known but significant chapter in the history of American Indians, and especially for that of the Indian Territory and Oklahoma. This instrument was fashioned in 1870, and contemplated at further annual joint tribal meetings mandated by the federal government following the Civil War. The Five Civilized Tribes — i.e., the Cherokee, Chickasaw, Choctaw, Creek, and Seminole — had been removed from the southeastern United States in the first half of the nineteenth century to lands obtained through the Louisiana Purchase. At the beginning of the American Civil War, these and other tribes of the Territory consummated nine treaties with the Confederate States of America (CSA). Following that conflict and those alliances, the tribes found themselves in a difficult position with the federal government: in a series of seven stipulations presented at Fort Smith in 1865 and in subsequent treaties in 1866, the federal government dictated the future for these peoples. Proviso number six from 1865 stated that "[i]t is the policy of the government, unless other arrangements be made, that all the nations and tribes in Indian territory be formed into one consolidated government, after the plan proposed by the Senate of the United States, in a bill for organizing the Indian territory" (Message of the President of the United States, and accompanying documents, to the two Houses of Congress, at the commencement of the first session of the Thirty-ninth Congress, 1866, p. 503).

A model for tribal meetings to expedite the development of such a formalized government for the Indian Territory was outlined in the provisions of Article 12 of the Treaty with the Cherokee, 1866 (Kappler, 1904, pp. 945-946): "The Cherokees agree that a general council, consisting of delegates elected by each nation or tribe lawfully residing within the Indian Territory, may be annually convened in said Territory, which council shall be organized in such manner and possess such powers as hereinafter prescribed." Six separate provisions in that Article defined the expectations of the federal government, and a series between July 1870 and March 1875 of six modest appropriations bills for $61,500 was passed to fund these activities. ^[1]

The tribes, after much delay, reacted to those 1865 and 1866 federal requests as one means to protect themselves against the impending avalanche of settlers and political changes (Burton, 1995, pp. 26-27). Nine council meetings took place in the Creek capital of the Territory during September and December of 1870; June of 1871 and 1872; May and December of 1873; May 1874; and May and September of 1875. Their task was to construct the basis of a single federal state from a number of sovereign Indian nations, an unprecedented opportunity. Table IA | Download Excel File displays the range of delegate tribes, taken from the Journal issues, in attendance at each of the nine sessions. An additional column in Table IA | Download Excel File identifies, for reference, each tribe's page number in Wright's A Guide to the Indian Tribes of Oklahoma (1951).

These few sessions were thus motivated by the tribes' desire to acquire a secure final homeland and by federal enthusiasm for these groups to postulate a constitution for a "Territory of Oklahoma" (see Treaty with the Choctaw and Chickasaw, 1866 ; Kappler, 1904, p. 922) in preparation for that forthcoming Indian statehood. The proposed entity's name reflected the tribes' total commitment to such an outcome: Oklahoma was derived from a "combination of two Choctaw words 'Okla' meaning people and 'Humma' red" (Origin of county names in Oklahoma, 1924, p. 80). The name was suggested by Reverend Allen Wright of the Choctaw Nation, who acted as a Commissioner for the federal government during the negotiations of the Treaty with the Choctaw and Chickasaw, 1866 that proposed that the Superintendent of Indian Affairs would serve as the Governor of this new Territory (Wright, 1936, p. 156).

During the December 1870 event, a formal constitution was promptly created, with slight revisions implemented at the May 1875 gathering. The document was crafted along the lines of other tribal constitutions, with direct or at least strong reference to the 1839 Cherokee, 1860 Choctaw, 1867 Creek, and 1867 Chickasaw constitutions in particular. ^[2] Altogether, there were 41 Chickasaw, 40 Choctaw, 38 Cherokee, and 18 Creek passages that were likely considered to direct the weaving of the final 1870 Okmulgee wording. It is apparent that the federal and CSA Constitutions — which were highly correlated themselves — were consulted (see the Appendices in Bernholz, 2011 for more on these transfers). Hill (1910, p. 487) called the Okmulgee instrument "the first practical plan for the government of the Indian Territory."

Okmulgee Constitution of 1870

The tribes collated their experiences to create the Okmulgee Constitution (see the text in Appendix I), but their own individual documents reflected intratribal awarenesses or concerns that were lost in the final fabric of Okmulgee. Indeed, it would have been nothing short of negligence if the Okmulgee Council participants had disregarded previous constitutions — federal, state, or tribal — during their planning of their proposal. Regardless of the contemporary newspaper reports in the mid-West and East that offered hyperbole when describing Indians beyond the Mississippi River, the Okmulgee Constitution established that the tribes took their task and responsibility very seriously (see Williams, 1879); that their constitutional product was a sound prototype, based on acceptable constitutional parameters; and that it clearly would be perceived as a manifestation of their needs and dreams for a better future. Evidence of the influence of a previous state constitution, as Burton (1995, p. 74) had referenced with regard to the development of these four earlier Indian instruments, is readily available: Article 11 of the 1796 Tennessee Constitution , for example, pertained to that instrument's Declaration of Rights and §6 stated a matching outcome to that found in Okmulgee: "That the right of trial by jury shall remain inviolate" (Poore, 1877, p. 1674).

The administrative responsibilities to organize productive Okmulgee Council meetings were substantial, given the multiplicity of tribal differences and their histories. The growing list of interpreters — from six to eleven members — between the December 1873 adjourned fourth and the May 1874 fifth Council sessions illuminated the ensuing complexity of these proceedings (Journal of the [Adjourned Session of the] Fourth Annual Session of the General Council of the Indian Territory, 1874, p. 3 and Journal of the Fifth Annual Session of the General Council of the Indian Territory, 1874, p. 4). Further, the efforts made to include the "wild" tribes in the Territory's future considerations were clearly a combination of a manifestation of deeply felt tribal internationalism with compliance with federal demands specified in the second Fort Smith stipulation — "Those settled in Indian territory must bind themselves, when called upon by the government, to aid in compelling the Indians of the plains to maintain peaceful relations with each other, with Indians in the territory, and with the United States" (Message of the President of the United States, and accompanying documents, to the two Houses of Congress, at the commencement of the first session of the Thirty-ninth Congress, 1866, p. 502). The inclusion of remarks, made by a wide variety of delegates and found in the later Journals for the fifth and the sixth annual sessions, was an expression of concern for enhanced participation by all tribes in the Indian Territory.

This dream of an Indian state actually died slowly between 1866 and the turn of the twentieth century, absent all the efforts extended by the tribes of the Territory; Buck (1907) highlighted many of the difficulties that arose between those in favor of, and those against, Indian statehood. Abel (1908, p. 101) provided statements, beginning in the year 1868, from the files of the Southern Superintendency that decried the delay in the formation of the General Council specified in the 1866 treaties, ^[3] the crucial organizational element that was assigned the responsibility of the eventual creation of the Okmulgee Constitution.

Nevertheless, what is critical to an understanding of the Okmulgee Constitution is the boldness and aptitude with which the Cherokee, Chickasaw, Choctaw, and Creek tribes had initiated diplomacy, formulated pacts with neighboring tribes, created functional constitutions, and instigated progress well before the post-Civil War demands espoused by the federal government. Indeed, in the case of the Cherokee, four decades separated their 1827 constitution — a document carefully modeled after the federal instrument — from the Okmulgee proceedings. That initial Cherokee perspective had been formed in the very teeth of early yet intense removal activities, when the tribe was then fighting (and had been so since the first hints of potential removal in 1809) to remain in the Southeast. This profound step by the Cherokee — and the inability of the other three tribes to make such a smooth transition over the next several decades — induced for these entities various delays in achieving badly needed institutional adaptations. In fact, the Choctaw had prepared a constitution prior to that of the Cherokee, but no substantial adjustment in functional organization was accomplished (Champagne, 1992, p. 1).

In the Message of the President of the United States communicating a copy of the proceedings of the council of Indian tribes held at Ocmulgee, in December, 1870 (1871, p. 1), President Grant declared his views on establishing for the tribes "forms of territorial government compatible with the Constitution of the United States and with previous custom toward communities lying outside of State limits." The supporting documents in that Senate Executive Report substantiated the efforts expended at the Okmulgee Council meeting, but the President's initial remark quickly indicated difficulties of situational perception. As Applen (1971) correctly titled his article, the tribes were under the distinct impression (both psychological and legal) that there was to be a direct transition between the current Indian Territory and a full-blown state reserved to the tribes. Undoubtedly, from the very beginning, the tribes operated under the idea that the federal government wanted them to fashion an entity devoted to the needs of the tribes, and that their council meetings, specified in no uncertain terms by the federal government after the Civil War and commencing with the June 1870 session, were to culminate in nothing less than that achievement.

Comparing text variants

The universe of text analysis relies heavily upon the ability to discern differences among variants of selected text. For some writers, constant tinkering with their material yields an abundance of evolving prototypes that propel such later analyses; the works of Walt Whitman may serve as an appropriate model of this behavior. ^[4] Bernard Cerquiglini's remark that "medieval writing does not produce variants; it is variance" (1999, pp. 77-78) should be declared the general motto underlying all text analyses that wish to ferret out the alleged isotopes (Higgins, 1997) of literature. Variability, though, is not reserved exclusively to medieval texts, but rather it acts as an incentive, as well as a cause, for sustained mining of all forms of materials. Even more mundane manuscripts — for treaties between sovereigns or for constitutions destined to guide such entities — will yield their own rewards, if tested. It is the variance, not the rigidity, which give these texts a life of their own. Additionally, their hidden richness makes an examination of the Okmulgee Constitution variants so much more attractive. The forces of legal strictness; unknown futures; and social diversity all combined to saturate this instrument with its own special mix of psychosocial components.

Text analysis tool selection

There exists a particularly useful device with which to address these kinds of text analyses. Vladimir Levenshtein, the 2006 Institute of Electrical and Electronics Engineers Richard W. Hamming Medal winner "for contributions to the theory of error-correcting codes and information theory," proposed in 1966 an algorithm to assess information transfer, where the three operations of deletion, insertion, and substitution may be engaged to correct errors contained in a transmitted string. To demonstrate its applicability, Soukoreff and MacKenzie (2001) used the two string models quick brown fox and quixck brwn fox as prototypic examples of presented and transcribed texts in such a task. While as many as six individual errors may be present in this communication — established by the failure of the xck br substring to accurately convey the initial ck bro material — the two most likely errors may be considered to be the insertion of the character x and the omission of the character o. As a result, these discrepancies yield a computed Levenshtein's edit distance (LED) score of 2 for this test, or for the total number of correcting operations required to first delete the x, and then to accomplish the insertion of the o. Identical strings — quick brown fox and quick brown fox — would require no corrective operations and would thereby produce a computed LED score of zero. Further, any observed LED must be less than or equal to the maximum length of the two strings, since replacing an entirely missing sequence with one of length n would require no more than n operations: here, quick brown fox vs. _____ _____ ___. The Levenshtein algorithm, thus, is very adaptable and has served in many diverse applications, including vehicle travel time measurement scenarios (Takahashi and Izumi, 2006) and the development of ontologies (Ginsca and Iftene, 2010), as well as the foundation of spell checking (Kukich, 1992) and plagiarism software (Zini, Fabbri, Moneglia, and Panunzi, 2006). However, in text analyses, these LED scores are particularly intuitive, since any string comparison that supports an LED of zero means complete similarity between the elements in question, while any non-zero returned value immediately identifies disparities and the magnitude of such differences.

These LED calculations may be made at two levels. First, they may be computed at the token or element level. Two relevant examples are apparent in the test involving the terms dog and dig with its LED score of 1, and in the cumulative score of 14 that is generated in an evaluation of just the first line of Walt Whitman's 1855 original and of his later 1891 revision of I Sing the Body Electric, i.e., for an assessment that evaluates the dissimilarity between "The bodies of men and women engirth me and I engirth them " and "The armies of those I love engirth me and I engirth them ." In this last scenario, the italicized words mark the four pairs of tokens that induce that LED score of 14 across these twelve elements.

A second option may be administered to ascertain disparities between entire documents. In theory, this latter approach might facilitate an examination of various editions of a specific author's work, or of a collation of statutes, or of similar ensembles. One advantage of such multiple comparisons occurs whenever the computed cumulative LED amount increases, decreases, or remains the same for later copies of the same material. Hypotheses relative to these three possible outcomes may stimulate the formation of a simple set of assumptions of the provenance of the test materials. Three possible suppositions for these three theoretical LED conditions suggest that they could be due to a) a lack of editing expertise across the suite of documents that grew worse over time, as evidenced by the accumulation of errors reflected in an increasing cumulative LED value; b) there might have been some degree of editorial intervention that corrected some resident errors and so the LED number automatically diminished over subsequent versions; or c) the constant LED quantity simply signaled that no apparent intercession occurred between the text of the first and the creation of the second (and/or later) rendition(s) and that the subsequent one(s) was a mere facsimile of the initial form.

The flexibility found within the realm of Whitman's poetry, however, is absent — or is nearly so — from legal contracts and treaties. Expressions of variability among alleged replications of these formats have led to endless bouts of litigation. ^[5] Indeed, in a previous analysis of the Treaty of Fort Laramie with Sioux, etc., 1851 (Bernholz and Pytlik Zillig, 2009 and 2011), the published federal texts for that instrument were interrogated in order to produce a final, correct version of the transaction, incorporating both the original document's material and a later amendment made by the Senate to one of its articles. As one consequence of that study, it was concluded that Fort Laramie had never been published in an entirely error-free state. ^[6]

LED score sensitivity and the Okmulgee Constitution test suite

Just as in the Fort Laramie study, the test data in this Okmulgee Constitution examination consisted of a vertically aligned joint set of the various published texts, stripped of delimiters and constructed to a uniform length, where any single document's alignment was augmented if needed by blank pad elements to fill in any absent subsection(s) of that version. As an example of this adaptation, two parallel texts might contain the terms two-thirds vs. two thirds which, when placed in these vertical arrays, would require that their element sequences occupy in the first instance just one location, but then two places for the second. A blank pad in the former corresponds to the latter's thirds term and thereby aligns the two text segments. The fundamental, or base, document — the two-thirds one here — may require padding in order to incorporate text styles from later interpretations, or vice versa. The Territory variant induces this modification at line number 879 and 880 of the data in Table IIA | Download Excel File, relative to the Council standard text:

	Council	Territory
879	two-thirds	two
880		thirds

The application of the LED algorithm to these data would compute two edit distance scores, one for each of the two-thirds vs. two and of the [blank] vs. thirds rows, and would return individual values of 7 and 6, respectively, for a cumulative LED of 13. Similarly, two variants with the terms to wit and to-wit (see 1041 and 1042 in Table IIA) would initiate an LED score composed of 4 plus 3 changes, or a cumulative score of 7. Note that a contrast between two-thirds and two thirds (or to wit vs. to-wit) as two strings instead of two pairs of elements would generate in each instance a cumulative LED of just 1, i.e., through the single insertion cost of the hyphen separating the two words in each of the target pairs. The vertical text distribution format employed here thus maximized these potential cumulative LED scores; the process was very sensitive to disparities. In addition, Levenshtein's process as designed returns evidence of all text differences, including those of capitalization, but since the main objective of that earlier study was concerned with the contents of the Fort Laramie treaty rather than with their presentation or format, all materials were first normalized to lower case prior to similarity testing in order to reduce unnecessary background noise in these calculations. These assessment conditions were replicated for the Okmulgee Constitutional data. Document titles were included as part of each file's data.

For the assessment conducted here, fourteen documents — contained in thirteen publications — were considered. These consisted of three primary documents, i.e., the Okmulgee Constitution created by the General Council in December 1870 and two versions of the revised instrument from September 1875; one Senate bill text that cited the material; five additional federal representations, three popular press reports, and one compendium example of the original Constitution; plus one other federal rendition of the revision. They are listed below, and the bracketed names are used as brief identifiers within the remaining text and Tables. The "1870 Constitution" worksheet in Table IIA | Download Excel File contains the first series of data, and the "1875 Revision" one in Table IIB | Download Excel File displays the later set.

Texts providing the original December 1870 document (N = 11)

• Journal of the General Council of the Indian Territory (1871) — 20 December 1870 — [Council];
• Indian Territory, Oklahoma (1871) — 19 January 1871 — [Territory];
• A bill to ratify and carry into effect the constitution and form of government for the Indian Territory adopted December twenty, anno Domini eighteen hundred and seventy, at Okmulgee, by the general council of said Territory, held by authority of the Government of the United States [20 and 25 January 1871] (1871c) — 25 January 1871 — [HarlanB]. Note that this is the amended version of Senate bill number 1237 that was introduced on 20 January 1871 by Senator James Harlan (R-IA). ^[7] That earlier document is denoted as HarlanA (A bill to ratify and carry into effect the constitution and form of government for the Indian Territory adopted December twenty, anno Domini eighteen hundred and seventy, at Okmulgee, by the general council of said Territory, held by authority of the Government of the United States [20 January 1871], 1871b). A third bill was created on 9 March 1871 as Senate bill number 80 and is named HarlanC in a series of subtests within this analysis (A bill to ratify and carry into effect the constitution and form of government for the Indian Territory adopted December twentieth, anno Domini eighteen hundred and seventy, at Okmulgee, by the general council of said Territory, held by authority of the Government of the United States, 1871a). The text of HarlanC is not presented with the other variants, but is used in a small inquiry comparing these legislative materials;
• Message of the President of the United States, communicating a copy of the proceedings of the council of Indian tribes held at Ocmulgee, in December, 1870 (1871) — 30 January 1871 — [Ocmulgee];
• Message of the President of the United States, communicating the second annual report of the Board of Indian Commissioners (1871) — 10 February 1871 — [Commissioners];
• Investigation of Indian frauds (1873) — 3 March 1873 — [Frauds];
• Constitution of the Indian Territory (1873a and b) — 21 and 28 June 1873 — [Vindicator];
• In the Senate of the United States (1879, pp. 613-620) — 11 February 1879 — [SenateA];
• Okmulgee Constitution (1925) — September 1925 — [Chronicles]. This text is available on the World Wide Web, through the Oklahoma State University Library's Electronic Publishing Center. The document is prefaced by a useful introduction, but the deployment in the Oklahoma State text of the error notation [sic], to identify twelve alleged misspellings, was not implemented in this study;
• Beckett (1930) — 1930 — [Beckett]; and
• Wilkins (2009) — 2009 — [Wilkins].

Texts presenting the revised September 1875 document (N = 3)

• Journal of the Sixth Session of the General Council of the Indian Territory (1875) — 15 May 1875 — [Sixth];
• Journal of the Adjourned Session of the Sixth General Council of the Indian Territory (1875) — 9 September 1875 — [Adjourned]; and
• In the Senate of the United States (1879, pp. 620-627) — 11 February 1879 — [SenateB].

The chronological order of these materials is indicated, with the following provisos. First, for the Journal of the General Council of the Indian Territory, the Journal of the Sixth Session of the General Council of the Indian Territory, and the Journal of the Adjourned Session of the Sixth General Council of the Indian Territory sessions, the final meeting date was selected for these conferences. Second, it is now impossible to confirm that the initial Okmulgee Constitution, presented in the Journal of the General Council of the Indian Territory (1871), was printed prior to any of the other 1871 federal materials. The same concern affected the three sources of the revised, September 1875 instrument (i.e., in the Journal of the Sixth Session of the General Council of the Indian Territory, the Journal of the Adjourned Session of the Sixth General Council of the Indian Territory, and the In the Senate of the United States publication), but in that case, the question of publishing order was mitigated by the range of dates. Third, the Constitution article from the Chronicles of Oklahoma was assigned the publication date of the journal's issue that contained this piece. Fourth, both the Beckett and the Wilkins entries were marked with their respective year of publication. Finally, Wilkins was a special case that was purposely published in a shortened version. LED analyses of this specific rendition will be discussed separately, since a clear provenance course may be proposed from the available text even though part of its full text was absent. Each of the Wilkins element cells in the "1870 Constitution" sheet of Table IIA | Download Excel File beyond line number 3595 has an em dash (—) to indicate that no physical data — and therefore no potential error — existed at this point. David Wilkins also included the apparent source for his Okmulgee Constitution rendition (2009, p. 134): the Message of the President of the United States, communicating a copy of the proceedings of the council of Indian tribes held at Ocmulgee, in December, 1870 (1871), i.e., the document that was identified in this study as Ocmulgee.

The variant list immediately highlighted a substantial chronological break between the nineteenth century materials and those full texts created in the twentieth century: both Chronicles and Beckett were, respectively, popular press items from a journal and a book dedicated to the history of Oklahoma. The Wilkins entry stood as an abbreviated entry in a recent compilation of pertinent documents that have affected the overall history of the tribes. All three of these versions were included in this analysis because, in many ways, they are probably the most easily accessible and, therefore, most frequently read accounts of the Okmulgee Constitution since its initial publication, either as an official document within the federal government (see SenateA and SenateB for both the original and the revised Constitution texts), or within local Indian Territory media (see Vindicator, published in the Choctaw Nation's newspaper of that name). ^[8] These subsequent renditions signaled a sustained interest in these affairs, not only in terms of the evolution of Indian Territory into the current state of Oklahoma, but also as a representative instrument that helped form and power the history of American Indians. Wilkins' Introduction statement made a particularly relevant observation regarding his overall compilation of these diverse documents. He said that "[t]he startling diversity evident in native governing structures, along with the value systems that underlay them, indicates that native peoples were more than willing to embrace new legal traditions and institutions in an effort to adjust to the shifting political, economic, legal, cultural, and vastly changed territorial conditions. At the same time, varying segments of many nations strove to maintain precontact sociocultural norms, institutions, and ceremonial traditions to distinguish themselves from other native nations and the inexorable tide of intruding powers" (2009, p. 2). Thus, this single tribal constitution may be thought to distribute more than just the words describing a vision for Indian independence on the Plains. It can reveal, as Wilkins proposed, the willingness of a number of tribes to adapt — almost without limits — and to form a meaningful confederation out of the chaos of removal, the Civil War, and the prospect of losing their lands, so that they as sovereign entities would not vanish.

Levenshtein edit distance calculations for the Okmulgee Constitution variants

The perpendicular arrangement of the variants induced alignment across individual elements within these texts. Introduced blank pads aided the rectangularization of the versions, since all of the presentations had disparate lengths (see the "Data characteristics" worksheet of Table IB | Download Excel File); these blank pads are purposely highlighted in Table IIA | Download Excel File and IIB | Download Excel File for all alternative texts. The use of such additions may be observed immediately in line number 1 of Table IIA | Download Excel File, where all renditions but Beckett required this intervention to address the sole appearance of the first element the in the latter's title. Similarly, the insertion of blank elements in line number 50 through 52 in each variant except HarlanB balanced the almost total utilization of the term 1866 with HarlanB's substitute application of eighteen hundred sixty-six. This inclusive standardization was the first step towards uniformly calculating the LED scores required to index the inherent dissimilarities among these texts.

Comparisons between pairs of columns — where Council served as the exemplar — yielded 3,866 individual token LED scores for each set of evaluations. The sum of these LEDs gave a cumulative score that expressed the magnitude of textual changes — here, calculated in bytes — necessary to bring that specific pair of vertical text arrays into register. A pair of identical texts would return zero errors and, therefore, set the lower bound of any cumulative LED at zero. The first two lines reported in Table IIIA | Download Excel File provide the number of errors observed and their cumulative byte value. This Table, for example, specified that there were 52 pairs of non-matching elements consisting of 216 bytes in the comparison of the Okmulgee Constitution as published in Council and Territory, but only 18 differences totaling 73 bytes in a test of the terms found in Council and HarlanB. The rectangular nature of these data allowed direct comparisons among the returned cumulative LED values, unencumbered by the need to readjust the findings based upon unequal text element totals. Thus, the observed error counts of 18 for the Council-HarlanB and 15 for the Council-Commissioners trials suggested similar accuracy in reproducing the Council text by each of HarlanB and Commissioners. The cumulative LED score of 73 vs. 46 bytes additionally proposed that, even with approximately the same number of faults, Commissioners was a marginally more accurate copy of Council than was HarlanB, based on byte accuracy.

The expected cost and expected benefit values assembled in Table IIIA | Download Excel File will be discussed below, but they may be understood as indicators of known textual problems and of known acceptable remedies, respectfully, that may be legitimately applied to diminish the overall net noise tally to create a true index of disparity between text arrays. The initial results from these Council comparisons stimulated other tests, e.g., an examination of the HarlanA vs. HarlanB was calculated to study the Okmulgee Constitution text presented in Congress within a few days (i.e., on 20 and 25 January 1871, respectively) as part of an initial and then as an amended bill, and published on each occurrence by the same federal printers. These special test outcomes were a window into that publishing domain.

The concatenation of all errors indicated that 299 individual text elements varied between one and ten times each across all variants, including the truncated Wilkins one. Table IIIB | Download Excel File shows the number of inaccuracies and the frequency of their occurrences. It also presents all individual discrepancies which, when multiplied by their error state counts, provides a grand total of 582 errors.

Initial observations

The Fort Laramie study indisputably set the foundation for this Okmulgee Constitution endeavor. It was not only a test bed for an evaluation of the Levenshtein metric, but provided insight regarding outcome possibilities. In general, two "families" of errors were observed that might be considered of major and of minor importance.

In the major category, comparison errors appeared either as text incursions or as text exclusions. In the Fort Laramie tests, the former shortfall was revealed in Kappler's 1929 version of the treaty, with a redundant section of sixteen words thence up the north fork of the Platte River to the forks of the Platte River in Article 5 (Kappler, 1929, p. 1066). The latter, omission fault was evident by passages absent from the 1873, 1903, 1904, and 1929 versions of the treaty that failed to include a nine-word phrase in the Article 5's boundary description for the Gros Ventre, Mandan, and Arikara territory that had been a part of the 1851 original text. The original document read thence up the Yellow Stone river to the mouth of Powder River; thence from the mouth of Powder River in a South-easterly direction to the head waters of the Little Missouri river, where the seven bold words were not replicated in any of those later 1873, 1903, 1904, or 1929 compilations. The surplus Platte River section may be seen in line number 353 through 368 of the 1929 rendition of the instrument, and the nonappearance of the Powder River sequence at line number 471 to 477 in the 1873, 1903, 1904, and 1929 documents in the composite Fort Laramie data . These significant departures from similarity provided guidance in determining the provenance of these various Fort Laramie presentations: the exclusion strongly tied together the 1873, 1903, 1904, and 1929 texts, while it simultaneously separated them from the set of 1851, 1852, and 1884 productions. The historical sequence was thus reinforced by these observations. The binding together of the 1903, 1904, and 1929 examples through these data was, in part, a reflection of their sequential production by Charles J, Kappler for successive second volumes in 1903 and 1904 of his Indian Affairs: Laws and Treaties ensemble that must have used as its primary source the material found in A Compilation of All the Treaties Between the United States and the Indian Tribes Now in Force as Laws (1873). The apparent linkage of the 1851, 1852, and 1884 texts was due to a certain extent by the Senate itself, which created the 1852 document as it considered the original, 1851 treaty transaction, while the 1884 became a relevant element in the third edition of a federal collection entitled Laws of the United States Relating to Indian Affairs: Compiled from the Revised Statutes of the United States enacted June 22, 1874, and from Statutes at Large from that date to March 4, 1883: Also, Special Acts and Resolutions Previous to the Enactment of the Revised Statutes, not Embraced in or Repealed by the Revision: Also, List of all Ratified Treaties and Agreements Made with the Several Indian Tribes (1884, pp. 317-319).

The second, minor class of errors included spelling discrepancies and the replacement of words. In Fort Laramie, the terms head chiefs, head men, fifty one, north westerly, south-easterly and south easterly, south-westerly, and Yellow Stone were part of the original treaty lexicon, while post-1851 documents delivered headmen, fifty-one, northwesterly, southeasterly, southwesterly, and Yellowstone. As noted above, the vertical arrangement of the data maximized the LED for these comparisons. In addition, the term alonge emerged in the 1873 reproduction (see line number 492 of the data set ), but in no other example. Here, the concept of the Levenshtein edit distance score as a benefit indicator prevailed: it would require one byte to delete that terminal e from alonge to make the necessary adjustment to align the 1871 version with the original treaty text. Such demonstrations were valuable for a more complete Okmulgee analysis.

Assignable costs in the Okmulgee Constitution

There were four major textual problems within the comparisons of variants of the December 1870 Council. Two had to do with individual nine word exclusions, one was the misspelling of the Council Secretary's surname, and the last was an instance of the complete absence of the text of the Schedule to the Constitution.

The first thirty byte exclusion — and on the south by the state of Texas — was defined here as the nine item, Texas boundary exclusion, and it was absent from the first sentence of Article 1, §1 in the Commissioners, Frauds, and Wilkins versions.

The second exclusion was forty-two bytes and nine elements in length — of having the witnesses to testify in his presence — and is referred to as the witness exclusion. It was missing from §6 of the Declaration of Rights in the Territory and the SenateA accounts.

At the end of the Council document, the surname for its Secretary, George Washington Grayson, appeared as Greyson. Four later Okmulgee publications corrected this error — i.e., in the Ocmulgee, Frauds, Chronicles, and Beckett alternatives — and even though this was but a single-byte disparity, its presence in the last name of such a prominent General Council participant was unfortunate.

Finally, the Wilkins rendition of the Okmulgee Constitution did not include the 270 term (including three pad elements) 1,309 byte Schedule to the Constitution. ^[9] This absence was similar to one observed in Kappler's 1903 version of the Treaty of Fort Laramie with Sioux, etc., 1851 (1903, pp. 440-42), in which the testimonium was removed as part of an effort to save space. ^[10]

Table IB | Download Excel File identifies the presence of these four specific text elements within the ten versions of the Okmulgee Constitution that were created subsequent to the initial General Council presentation; the latter is presented for reference.

Assignable benefits in the Okmulgee Constitution

This Okmulgee Constitution examination was founded in part upon a desire to learn the provenance of the original 1870 instrument, and the contents of the underlying text and its variants. This meant that observed discrepancies were considered as valuable data; their occurrences were not blindly defined as a fault or an error but rather as a manifestation of some difference among the versions. Such a perspective returned two series of observations: those that may be classified as denoting assignable costs, as mentioned above, and those that may be construed as assignable benefits. These latter were, in general, unexpected but their presence was an aid to the understanding of the condition of the base Council material. Here, the original 1870 instrument contained the terms formes, agains, thist, cammission, and organized (see line number 43, 1970, 1974, 1978, and 3604, respectively, in Table IIA | Download Excel File, instead of the likely elements forms, against, this, commission, and organize in the other ten renditions. A comparative test between that 1870 Council base document and any other account thus created an LED score that was amplified by a total of five bytes that, through their service, would induce these corrections. Indeed, the Territory text — written less than a month after the Constitution was created at the Council gathering at the Creek capital — appeared in the Congressional Globe with these five corrections already in place (Indian Territory, Oklahoma, 1871). These findings suggested that textual interrogation through the LED metric may be a potential calculator of benefit as well as cost in such studies, and an avenue towards a better understanding of editorial/printer intervention(s).

More specifically, these five spelling errors were distributed throughout all major portions of Council. The term formes (line number 43) appeared in the preamble (Whereas the people of the nations of Indians inhabiting the Indian Territory have agreed by treaty with the Government of the United States, and been by its agents invited to meet in General Council under the formes prescribed by the Treaties of 1866 and the action thereon of the Government of the United States....). ^[11] The formes incompatibility was reminiscent of the usage of Supintent in the same document portion of the Treaty of Fort Laramie with Sioux, etc., 1851. ^[12] The three terms agains, thist, and cammission (line numbers 1970, 1974, and 1978) were detected in §9 of Article 4 (The Governor may grant pardons, and respites and remit fines for offenses agains the laws of thist Territory, and shall cammission all officers who shall be appointed or elected to office under the laws of the Territory.). Finally, the element organized was inappropriate for the first line of the original Schedule to the Constitution (line number 3604): In order to organized the Government of the Indian Territory, and secure practical operation for the same, it is hereby ordained....

These five bytes stand alongside the absence of three bytes in SenateB, the text of the 1875 revised Constitution, when the entire specification for impeachment in §15 of Article 3 (see line number R-1451; emphasis added) was impaired by the phrase no person shall be convicted with the concurrence of two-thirds of the members present. Certainly, without was the required term in this declaration. The few spelling bytes contained in formes, agains, thist, cammission, and organized of the original 1870 Okmulgee Constitution were repaired in later versions, while its revised 1875 rendition, seen in SenateB, was weakened during preparations for inclusion in the Serial Set. The manipulation of those initial but incorrect terms ultimately served the desired outcome that is very much in line with Levenshtein's approach. The title of his paper — Binary codes capable of correcting deletions, insertions, and reversals (1966; emphasis added) — demonstrated that the ultimate goal must be corrective in nature, and not just an enumeration of element differences or, specifically, of severe textual errors, as found in the impeachment phrase in SenateB.

Implications of the exclusion table

Exclusions, by their very nature, make bold statements, especially in materials that are considered as, or are candidates to become, the law of the land. The Okmulgee Constitution was much more than a rough draft for the future. It was conceived as the basis of a tribal application to convince the federal government that the people of the Indian Territory were both prepared and adamant about an Indian state within the Union. There are, thus, a number of possible yet pertinent implications that may be derived from the exclusions posted in Table IIA | Download Excel File. In this consideration, HarlanB — an internal working document of Congress — was excluded from consideration:

It was immediately clear that the official copy of the Okmulgee Constitution in the Congressional Globe, i.e., the one called here the Territory version, was unable to convey the complete document, given the witness exclusion (line number 3330 to 3338) from the Declaration of Rights portion;

The Ocmulgee document published in Serial Set volume 1440 excluded the Texas boundary definition (line number 206 to 214) and thereby created an instrument with an ill-defined geographic range;

The first reliable reproduction of the original Okmulgee Constitution occurred in the 1871 Commissioners document of the same Serial Set volume, since Territory and Ocmulgee from earlier in that year had previously suffered from the witness and the Texas exclusions, respectively;

The Texas boundary exclusion found in the 1873 Frauds was apparently replicated directly from 1871 Ocmulgee, since these two publications were the only nineteenth century reproductions sharing this shortfall;

In 1879, the errors within the SenateA material directly announced that it was reproduced from the 1871 Territory account; these two alone share the witness exclusion; and

The 2009 Wilkins account was the most divergent from the original instrument, with both the Texas boundary specification and the entire Schedule of the Constitution missing, regardless of the editorial rationale for the latter.

LED testing and results

Table IIIA | Download Excel File tallied the discrepancies found in the comparisons between Council and each of the ten renditions of the 1870 Constitution, and of the match between Sixth and the two parallel versions of the revised document from 1875. In this display, WilkinsF is the name associated with a full-length test between all elements of Council and this variant, while WilkinsT identifies the results with the truncated text, i.e., for the analysis that excluded any LED calculations for the Schedule to the Constitution segment. There are, in total, 849 disparities in the original document tests and 59 in the revision tests (see the original data sets of Table IIA | Download Excel File and IIB | Download Excel File). However, since the analysis between Council and Wilkins was affected by the absence of the Schedule to the Constitution subdivision, which accounted for 267 tokens and three pads in the composite data table, its removal reduced the total number of true errors across all ten assessments to 582 elements. Further, duplicate faults among tests existed that inflated the error list by their multiple occurrences — the term formes, at line number 43 in Council, was the prototypic case that induced ten observed errors in LED assessments performed across the remaining renditions. A list of unique errors in all ten comparisons was therefore 299 elements long. Of these, 176 items (or almost 59%) were single-event mistakes: these may be seen in the misspelling of guarantied in Territory (line number 221); in the replacement by eighteen hundred and sixty-six in HarlanB at line number 49 to 52; and in the incursion of [that are] in Wilkins at line number 3008 and 3009.

Revelations from Okmulgee variant data

Comparative text analysis thrives on inconsistency. If variants do not emerge, the entire endeavor almost ceases to exist. The reasons for the observed nests of dissimilarity may be simple or complex, yet no single document ever seems to pivot exclusively upon a "simple" explanation. Far more rigid instruments, like treaties, have less potential flexibility because, as Aust (2007, p. 16) has observed, these pronouncements "are drafted according to standard forms and processed according to long-established procedures." Their status as the law of the land should instantly further insulate them from adjustment, cosmetic or otherwise. Nevertheless, even renditions of legal materials reveal inconsistencies, as illuminated by the recent Treaty of Fort Laramie with Sioux, etc., 1851 results (Bernholz and Pytlik Zillig, 2009). It was quickly apparent that the situation would be no different with the Okmulgee Constitution: the Okmulgee data tables mimicked all error formats found in the Laramie study.

It is critical to keep in mind that all faults are not created equal, and that their distinct level of severity has a range that almost mirrors that of their collective richness of divergence. The thence up the north fork of the Platte River to the forks of the Platte River incursion in the 1929 Laramie account was not a stream of sixteen randomly selected terms that was fortuitously deposited into a contract among sovereigns. Rather, it was an utterance that was very highly correlated with the surrounding document syntax, even in its unwarranted state. Thus, errors embedded in compared texts may be difficult to observe easily, other than through mechanical processes like those employed in the Laramie and Okmulgee studies that maximize their visibility. Indeed, the course of making such measurements in this precise manner amplified minute alterations: testing Yellow Stone and Yellowstone in Laramie in this manner — through allocation to an LED table as a pair of two terms, Yellow and Yellowstone, and Stone vs. [blank] — turned a single-byte incongruity into two trial scores of five and five, or ten total bytes. On occasion, an editorial decision to make clearer a variant of the Okmulgee Constitution induced new calculated Levenshtein debt. David Wilkins established bracketed terms to expedite this clarification effort, but his use of [to] at line number 438; of the suffix to create process[es] at line number 2643; and of two occurrences of [that are] at line number 3008 and 3009 and at 3574 and 3575, accounted for four, four, nine, and nine LED bytes, respectively, within this exercise. Note too, though, that an analogous editorial influence led to the correction of the five familiar elements formes, agains, thist, cammission, and organized from the initial publication of the Constitution. These now valid spellings cannot be blamed on subsequent inattentive or error prone printers, as many subsequent textual blunders appear to be. Thus, squeezing a large LED calculation from the almost invisible shortfall evident in the contrast between Yellow Stone and Yellowstone demands that such known expected costs are balanced, at least in part if possible, by any recognized expected benefits attending, say, later more appropriate spelling(s). Subtracting the total byte count of both the expected cost(s) and the expected benefit(s) from the overall cumulative LED from a pair of instruments means that the final net noise value linked to that comparison is a more valid index of the actual magnitude of the inaccuracies rooted in those passages. This maneuver — the acceptance of the quantification of benefits observed through the application of the Levenshtein metric — was a step forward in the assessment technique employed during this Okmulgee project. The Laramie study was designed primarily to return a more accurate final treaty text, based on the original 1851 transaction and modified solely as stipulated in the later Congressional annuity adjustment. The Okmulgee endeavor looked more to the fluctuations, more to the cumulative effect of the mistakes, yet simultaneously desired to give credit when due for instances of ensuing smart textual modifications.

The perception that this examination of the Okmulgee Constitution pivoted upon a similar series of discrepancies as found for Laramie was assisted by the imposed data formatting. Each text was written in English; the expected legal formality or protocol was sustained during reproduction; and the overall relative orderliness induced by the more rigid alignment conditions minimized textual instability and eliminated the wild incongruities observed in other studies of more flexible literature.

Error forms

In text reproduction undertakings that sidestep translation and editorial intercession — and especially for those attempts in which the document possesses some legal or official weight and for which style or subjective interpretation is absent — the forms that errors may take are limited. These may include one or more faults resulting from misspelling, juxtaposition, replacement, exclusion, and/or incursion. Further, successive editions carry the opportunity to introduce more inaccuracies, so that later renditions should show, in general and regardless of the source(s), increasing numbers of divergences from the original. The variants of the Okmulgee Constitution in this study contained examples from these five classes of difficulties and they support the prediction of increasing fault creation over time. Of the 299 unique errors distributed throughout the Okmulgee variants, there were 103 identified as spelling mistakes; six determined to be juxtapositions; 111 replacements of the original 1870 material; 53 occasions of exclusion; and 26 intrusions of new text.

Misspelling (N = 103)

Misspellings are perhaps the most dominant and expected toll in the reproduction of texts. Plurals may be formed or missed (e.g., session vs. sessions and powers vs. power at Wilkins data line number 1607 and 3527, respectively); hyphenated words may be presented as separate terms absent the dash (two-thirds vs. two thirds at line number 879 and 880 in Territory); style differences may emerge (defense vs. defence at line number 961 and offenses vs. offences at line number 1969 of Vindicator); and/or general errors may prevail (approproiations at line number 1291 of Chronicles and whjch at line number 2040 of Vindicator).

Juxtaposition (N = 6)

In these rare instances, original word order was compromised and rearranged: see line number 347 and 348 for the phrase are hereinafter vs. herein after in Wilkins; the expression not have vs. have not at line number 1690 and 1691 for SenateA; and the terminology be twice vs. twice be residing at line number 3468 and 3469 of Chronicles and Beckett. This last error pair, in fact, served as one piece of evidence to recommend that Beckett was taken directly from Chronicles.

Replacement (N = 111)

Replacement involved the physical substitution of one word for another. For this study, errors were placed in this category if the new word was beyond a clear case of misspelling. The most extreme example in these Okmulgee data was the conversion of the year 1866 to the string eighteen hundred and sixty-six at line number 49 to 52 and 2440 to 2443 of HarlanB. Many of these adjustments made use of abbreviations for the terms article or section; there are six such paired transitions for article in Chronicles and Beckett at line number 162, 460, 523, 1497, 2388, and 2943. Similarly, the element sec in Council was changed to section 78 times in Territory, Ocmulgee, Commissioners, Frauds, SenateA, and Wilkins. More hidden adaptations included the deployment of being instead of been; of & instead of and; of office in place of service; and of the rather than a at line number 31, 649, 1527, and 2603, respectively, in Chronicles and Beckett exclusively. These findings too supplement the conviction that Beckett was derived from Chronicles. SenateA used Arabic, and not Roman, numerals to identify Okmulgee's six articles (see line number 163, 461, 524, 1498, 2389, and 2944), so there are shades to the absolute impact of replacements.

Exclusion (N = 53)

Exclusions addressed text segments that have disappeared, relative to the original 1870 Okmulgee declaration. Thus, an exclusion was revealed whenever gaps emerged in the composite data table; these voids signaled some sort of inability during reproduction to replicate faithfully the original. Their presence was especially useful to postulate links connecting text cousins, or between subsequent documents that might have shared a true, common predecessor.

The main exclusion in these data took place in Ocmulgee, Frauds, and Wilkins at line number 206 to 214, where the nine term Texas boundary definition and on the south by the state of Texas was absent. This deficiency was extremely strong proof that Frauds was derived from Ocmulgee, in a manner similar to that for Wilkins; David Wilkins actually included Ocmulgee in his bibliography as the source for that transaction. An additional reinforcement for this conjecture may be derived from the noted printed position of the Schedule to the Constitution that preceded the instrument text and Declaration of Rights in Ocmulgee and was reproduced in the same manner for Frauds. Wilkins remarked that he had not included the Schedule in Wilkins because of page restraints imposed by his publisher (David Wilkins, personal communication, 24 June 2010), but there remains the possibility that the Schedule was either skipped or deemed unimportant during the Okmulgee Constitution text accumulation phase of his writing.

Incursion (N = 26)

Incursions were represented by new material introduced into the primary text, in a complementary process to exclusion. These effects differ from replacement, since incursions are nested within the initial text, while replacements change the current wording. In the former case, the LED vertical arrangement of the primary (and possible other) material must be padded to make room for the new incursive element(s). The replacement process simply exchanged tokens, such as the substitution of office for service evident in this study at line number 1527.

Any injection of new matter into the mass of original elements was viewed as an instance of incursion. The first specimen occurred in the Beckett rendition at line number 1, with the immediate use of the definite article the to precede the recognized general title of the Constitution, but there was also the word that placed at line number 605 only in Commissioners, and Wilkins added the bracketed terms [to] at line number 438 and [that are] at line number 3008 and 3009 as well as at line number 3574 and 3575. These modifications may have affected the readability to some positive degree, but they were incursions nonetheless. Overall, there was no substantive equivalent in Okmulgee to the sixteen word incursion thence up the north fork of the Platte River to the forks of the Platte River from Article 5 of the Treaty of Fort Laramie with Sioux, etc., 1851.

Leverage from the Okmulgee error table

Table IIIB | Download Excel File identifies the 299 unique token differences found in the comparisons of the Okmulgee variants, grouped by error occurrence count. These reflect all classes of faults — i.e., those based on misspelling (marked as S), juxtaposition (J), replacement (R), exclusion (E), and incursion (I). In addition, one mistake was highlighted at line number 1829 of the three-error count group; three at line number 3860 to 3862 in the five problem set; and one for the eight occurrence group at line number 3858. These five special cases exhibit multiple difficulties for a specific token: emolument was misspelled as emoluments as well as replaced by employment; Indian affairs president was either excluded or abbreviated as Ind affs pres; and supt was either replaced by superintendent or excluded. In the first instance, the mistake was assessed as a replacement, while the others were identified as exclusions, since replacement and exclusion were deemed more severe processes than misspelling and replacement, respectively. The separation of hyphenated words was seen as paired element lines (see line number 670 and 671 for bona-fide; line number 879 and 880, 1379 and 1380, 2059 and 2060, 2092 and 2093, and 3712 and 3713 for two-thirds; line number 1041 and 1042 for to-wit; line number 1148 and 1149 for per-diem; line number 2335 and 2336 for attorney-general; line number 2958 and 2959 for three-fourths; and line number 3660 and 3661 for duly-authenticated). The bifurcated example at line number 3014 and 3015 contained a second error caused by the misspelling three-fourth in line number 3014 of Chronicles. As a result, these two token lines do not appear together within the single-error subset.

Reflection upon these and on the rest of the ensemble of error elements created opportunities to develop a better understanding of the provenance of presentations made subsequent to the initial Constitution in December 1870. After all, these variants were created to be considered as identical, or nearly identical, representatives. This assessment formed a necessary second and parallel endeavor to the survey of earlier tribal constitutions that was originally undertaken to develop a more formal understanding of the basis for the legal underpinnings of this 1870 instrument, and of whether textual identicalness was sustained. Just as segments of older constitutional expressions were carried forward into the post-Civil War instruments created by the same tribes in Indian Territory, the substance of those documents was reproduced (albeit, with errors) in later copies. This Okmulgee investigation, however, included newspaper and popular press items that, while not immune to the influence of federal documents, were produced nevertheless at a much greater geographical distance from those resources than the other related governmental items. The text found in Vindicator, Chronicles, and Beckett deserved special attention, since these were purveyors of more local historical description. The linkage between Chronicles and Beckett has already been mentioned: the Chronicles and Beckett versions provided only the Secretary's name — and expelled the seven elements of Enoch Hoag's name and position — at the conclusion of the Schedule to the Constitution (see line number 3856 to 3862); there were six paired transitions of article into art only in Chronicles and Beckett; there was the presence of the term being for been; of & instead of and; of office in place of service; and of the rather than a within Chronicles and Beckett exclusively. The correct spelling of George Washington Grayson's name in Ocmulgee, Frauds, Chronicles, and Beckett augmented the probability that Chronicles was a derivative of Ocmulgee and/or Frauds.

Frauds was produced two years after Ocmulgee, so the provenance of Grayson's correctly spelled name may have originated in Ocmulgee, the report that directly recounted the events as observed by three members of the Board of Indian Commissioners. Slight spelling differences — two-thirds vs. two thirds at line number 879 and 880, 1379 and 1380, and 2059 and 2060; three-fourths vs. three fourths at line number 2958 and 2959 — suggested that Ocmulgee was closer to Council than Territory, but the presence of punishments and Grayson coupled with the absence of the phrase Indian affairs president (at line number 3430, 3865, and 3860 to 3862, respectively) were indications of a more independent creation, as discussed earlier for Ocmulgee.

Territory, HarlanB, and Ocmulgee were all created during the month following the signing of the Okmulgee Constitution, within a span of just eleven days according to the documents' dates. There may have been two fundamental versions of the Constitution used to address later needs: the first should have been the initial Council variant, taken directly from the Constitution printed by Excelsior Book and Job Printing in Lawrence, Kansas, and a second from Ocmulgee, the federal document recollecting the visit to the Council session by the Board of Indian Commissioners. The almost unfathomable difficulty with Greyson vs. Grayson was especially blatant and this lent credence to the chance of two textual options: a collection of Council, Territory, Commissioners, Vindicator, and SenateA that displayed the first, incorrectly spelled surname of the Secretary that was thereby divergent as a group from a document set consisting of Ocmulgee, Frauds, Chronicles, and Beckett in which the name was spelled appropriately.

An extra advantage of the Levenshtein edit distance metric was that the number of observed mismatch errors gave an immediate indication of the strength of such document clusters. Table IIIA | Download Excel File revealed that the Levenshtein tests between Council and each of Commissioners, Frauds, Vindicator, SenateA, Chronicles, and Beckett generated 15, 57, 33, 50, 80, and 81 errors. Comparisons between Ocmulgee and these six yielded mistake counts of 46, 9, 70, 81, 78, and 79. Thus, only the results linking Ocmulgee with Frauds would recommend that Ocmulgee, instead of Council, was the source for Frauds. ^[13] The accompanying, supplemental material for the resolution pertaining to the Plains tribes, found in the Serial Set volume, may also be considered as verification of this union.

These returns provide more insight, however. It is apparent that misspellings occur with some frequency among these variants, but that exclusions are quite rare: the Texas boundary, witness, and Hoag exclusions are the prime examples in this survey. Their existence offered far more strength to a declaration of document similarity than did any series of single word faults. In the case of Council vs. Ocmulgee, there was an occasion to demonstrate this effect. Even though Ocmulgee, Frauds, Chronicles, and Beckett shared the Hoag exclusion at line number 3856 to 3862, only the first two renditions additionally suffered the Texas boundary exclusion at line number 206 to 214. Thus, Chronicles and Beckett must have been formed from a combination of previous editions and not just from a single foundation. The obvious downside to that blend was the very high error count and cumulative LED scores for each of the Chronicles and Beckett tests against Council.

Much like the Ocmulgee-Frauds pairing, an LED test for Chronicles-Beckett discovered only 29 errors and a total of 44 bytes of dissimilarity. A parallel situation arose with Territory and SenateA. These two exhibited the nine element witness exclusion at line number 3330 to 3338 — of having the witnesses to testify in his presence — from §6 of the Declaration of Rights. Here, the tests between Council and these two other editions of the Okmulgee Constitution caused 52 and 50 errors, respectively, but in a test between the Territory and SenateA, only 44 errors composed of 197 bytes were obtained. A tighter fit might have been expected if the exclusion hypothesis had been in effect for this set of texts, but twelve of the 44 errors were due to six pairs of changes to article notations (i.e., Roman numerals replaced Arabic ones) and to the conversion of section to sec; the latter accounting for 24 bytes, or about an eighth, of the cumulative LED. Thus, even in the event where a single source is under reproduction, editorial intervention and/or style modifications can overwhelm a clear view back to that original document. The relatively abundant noise in the Chronicles and the Beckett variants was a strong index of this impediment.

The revised Okmulgee Constitution of September 1875

Delving into an inquiry of the revised Okmulgee Constitution was a procedure for understanding the demise of the entire constitutional effort. Six months after the creation of the Constitution, and at the second annual conference, it was revealed that the Creek had stormed ahead and already ratified the proposal, but other tribes had not (Journal of the Second Annual Session of the General Council of the Indian Territory, 1871, p. 7). A provisional government was therefore proposed, in order to get a functional administration underway by mid-1872, even if forced elections to determine officials were considered the only viable alternative (pp. 11-12). This collapsing conviction was evident a year later at the third annual Council session in 1872, when it was announced on 6 June that just the Choctaw, Creek, Eastern Shawnee, Ottawa, Peoria, Quapaw, Sac and Fox, Seneca, and Wyandot had ratified the 1870 Constitution (Journal of the Third Annual Session of the General Council of the Indian Territory, 1872, p. 7). This was some progress in the desired direction, but the news was accompanied by the remark that "[t]he action of some of the tribes has not yet be ascertained," which made it even more obvious to all delegates that the two-thirds voting threshold required to confirm ratification of the Constitution had not been reached during the previous year and a half. Rather, dissention was already present at this meeting in 1872, illustrated by the formation of a quorum fully eight days after the planned onset of the conference (p. 9), and by a resolution that "a committee of five be appointed by the acting President, to report what measures, if any, are necessary to compel the attendance of absent members" (p. 8; emphasis added). The Cherokee were blatantly unenthusiastic about certifying Okmulgee because of the future place within the organizational structure allocated to them under the proposed constitution; the Chickasaw had met immediately after the end of the 1872 session and had "rejected overwhelmingly" the ratification proposal because of a perceived problem with equal representation; the Choctaw were adamantly against accepting land in severalty and only confirmed the instrument as a mode of insurance to protect their future interests; and the Seminoles balked at committing to the contents of a document and its listed processes that they did not fully understand (Nolen, 1980, pp. 274-275).

Protests against various federal activities were sent to Congress and the President, but the tide had already turned in Washington and the prospect of an Indian state evolving from Indian Territory was truly slipping away. Even under these circumstances, there were several more Okmulgee Council sessions and at the fourth such event in May 1873, a brief resolution was submitted to revisit the Constitution (Journal of the Fourth Annual Session of the General Council of the Indian Territory, 1873, pp. 10-11). The committee specifically assigned this revision task offered its own modified preamble and proposals that might finally resolve the impediment caused by the lack of tribal responses to the ratification request. It was, in part, a desperate attempt to stimulate all those involved to move forward as the situation progressively deteriorated, but at the general meeting convened in December 1873, the final blow was struck: the Clerk of the meeting "reported that none of the nations to whom the Constitution has been submitted, had reported action thereon" (Journal of the [Adjourned Session of the] Fourth Annual Session of the General Council of the Indian Territory, 1873, p. 9). In actuality, only 48% of the Indian Territory participants had cast their lot in favor of ratification; Appendix B of the Journal enumerated the individual tribal tallies (p. 21) and made apparent that a minority — just 32,065 of 66,461 voters — had supported the proposed Constitution.

Further council sessions were unproductive. At the fifth convention in May 1874, virtually nothing was said about the Constitution and the Journal for that year served as nothing more than a statement repository of the spokesmen from the Cherokee, Creek, Seminole, Eastern Shawnee, Confederated Peoria, Seneca, Wyandot, Ottawa, Sac and Fox, Delaware, Osage, Absent Shawnee, Wichita, Comanche, Waco, Caddo, Ionie (today, the Hainai), Pawnee, Keechie (Kichai), and Towoccanie (Tawakoni) (Journal of the Fifth Annual Session of the General Council of the Indian Territory, 1874, pp. 8-33). The sixth session began on 3 May 1875 and there was an immediate resolution passed for "re-submitting the Okmulgee Constitution to the president of the United States for his action" (Journal of the Sixth Annual Session of the General Council of the Indian Territory, 1875, p. 9), but Nolen (1980, p. 278) reemployed the phrase "a dead issue" to describe the delegates' perception of the 1870 Constitution at that moment. An adjourned special session, exclusively for an appointed constitutional committee, was set for 15 June 1875 "to draft a constitution," with the full Council destined to reconvene on 15 September (p. 72). A footnote on the very last page of that sixth annual session's Journal was initialed by "E. H." or Enoch Hoag, the President of the Okmulgee Council, and stated: "The foregoing draft of Constitution, prepared by a Special Committee of the General Council, is here published for the information of the delegates to said Council, who adjourned to meet in September next to act upon the same" (p. 114). In September, Hoag announced upon the acquisition of a quorum that "the businesss [sic] first in order would be the consideration of the report of the committee appointed to draft a constitution, by the previous session of the Council" (Journal of the Adjourned Session of the Sixth Annual General Council of the Indian Territory, 1875, p. 6). That report was provided by J. P. Folsom of the Choctaw: "We, your special committee, who were appointed pursuant to the recommendation made to and adopted by this Council in May 12th, 1875, whose duty was to prepare and perfect a draft of constitution to be submitted to the General Council of the Indian Territory for its consideration and action at this adjourned session, would beg leave to submit the following draft of constitution and ask for its adoption" (p. 7). His presentation was complemented by the panel's fresh constitutional attempt (pp. 8-20; titled here as Adjourned). The document was thereafter discussed and then — on the morning of the final day of the council — a motion was adopted "to postpone further action on the Constitution to the next session of the Council," i.e., to the first Wednesday in May 1876 (pp. 29-30).

That 1876 meeting never occurred, because the federal Indian Office terminated funding for such activities (Debo, 1934, p. 216, n. 122). Locally, The Vindicator newspaper announced on 26 April that "the Okmulgee Council will not convene again until further authorized by Congress" (Okmulgee Council, 1876). ^[14] The dreams of, and the struggle for, a true Indian state within the Union were, for all intents and purposes, over by that Spring. Based on this outcome, Applen (1971, p. 97) concluded that "[i]t is futile to speculate on what the future of the Indian territory might have been had the General Council succeeded in establishing an Indian government. When one considers the tenor of Congress during those years, and looks at the aggressiveness of the railroads, land speculators, and Kansas farmers, it seems unlikely that the Territory's later history could have been much different. The reason most often given for the failure of the proposed Indian government is that most of the Indians felt that their own intertribal government would eventually lead to United States territorial government. However, these same Indians apparently realized that an Indian 'state' was their only real hope for protection from further advances by the white men. Thus, their reasons for establishing an Indian government seem to have been just as strong as their reasons for not establishing one. In view of this, it is apparent that the old problem of representation was a significant, and probably a major, factor in the General Council's failure." In a parallel vein, Nolen (1980, p. 279) determined that "[i]t was the federal government which had urged the Indians to meet in general council, but it was the self-determination of the Indians that kept them from according to the government's wishes."

Those two texts of that revised Okmulgee Constitution — named here the Sixth and Adjourned — were joined three and a half years later by the document's republication in a Senate report (In the Senate of the United States, 1879, pp. 620-627; called now SenateB). ^[15] These three variants of the revised document — collected from two privately published pamphlets and from a Senate survey of railroad and Indian issues issued by the federal government — defined the suite for the next Levenshtein assessment. The instruments' tokens were assembled in the manner employed for the 1870 tests and renumbered with a unique line number prefix; here, "R-" for the revised document. These data appear on the "1875 Revision" worksheet of Table IIB | Download Excel File. The final length of the array, including seven pads, was 4066 elements. LED scores were computed for contrasts of the Sixth-Adjourned and the Sixth-SenateB relationships. Table IIIA | Download Excel File furnishes summary statistics that divulge that these trials revealed just 28 and 31 errors, and 106 and 78 bytes of dissimilarity, respectively. The measurements of likeness among the revised Constitution renditions were thus better than all comparable tests between Council and the reproductions of that earlier and shorter instrument, save for Commissioners and HarlanB.

As would be expected, the observed difficulties were nearly identical in format to those obtained in the Council contrasts. In the Sixth-Adjourned study, misspellings (such as o at line number R-1672 of Sixth in Table IIB | Download Excel File, the phrase the members vs. them embers at line number R-2131 and 2132, or convction vs. conviction at line number R-3526); juxtaposition (be neither vs. neither be at line number R-1873 and 1874); style (three-fourths vs. three fourths at line number R-3081 and 3081 and offences vs. offenses at line number R-2039); replacement (is vs. be at line number R-2918); and incursion (the elements at and is at line number R-3128 and R-3543, respectively, of Adjourned) were observed. Sixth-SenateB had parallel issues: misspellings (assented vs. asssented at line number R-50, criminal vs. crimina at line number R-2769, or convction vs. conviction at line number R-3526); style (first vs. 1st at line number R-150, section vs. sec at R-241, bona fide vs. bona-fide at line number R-783 and 784, and defence vs. defense at line number R-1040); replacement (for vs. of at line number R-104); and exclusion (the elements also and ever at line number R-2436 and R-3769, respectively, are absent from SenateB).

Potentially catastrophic errors

Among the collection of detected faults throughout all the texts involved in this study, the true meaning of the Okmulgee Constitution was never violated by any of these types of problems, other than through the difficulties imposed by the Texas boundary and the witness text exclusions. In those two situations, the boundary definition was absent from the first sentence of Article 1, §1 of Commissioners, Frauds, and Wilkins, and the witness specification did not appear in §6 of the Declaration of Rights of the Territory and the SenateA accounts. Other misspellings were unfortunate occurrences, but the use of two-thirds instead of two thirds, or of members instead of member, did not destroy the essential meaning of the passages. In SenateB, however, the situation changed drastically, based upon the misuse of a single word. The substance of Article III, §15 was written to provide guidance on impeachment proceedings before the House of Representatives and the Senate of the proposed Indian state. SenateB, one of the federal presentations reporting the 1875 revised Constitution, contained the following three sentences in its rendering of that section: "The House of Representatives shall have the sole power of impeaching. All impeachments shall be tried by the Senate. When sitting for that purpose, the Senators shall be on oath or affirmation and shall be presided over by the Chief Justice; and no person shall be convicted with the concurrence of two-thirds of the members present" (In the Senate of the United States, 1879, p. 622; emphasis added).

SenateA, within the same publication (p. 615; emphasis added), supplied the original 1870 Constitution text for the same Article II, §15 — "no person shall be convicted without the concurrence of two-thirds of the members present" — a condition that might be more predictable for the adjudication of such measures. Sections taken from the earlier 1839 Cherokee, the 1860 Choctaw, and the 1867 Chickasaw constitutions all spoke of without the concurrence of two-thirds of the members present. Thus, on the one hand, the LED calculations indicated a very successful attempt by Adjourned and SenateB to reproduce the Sixth's text of the revised Constitution. On the other hand, the Levenshtein quantification of mistakes between Sixth and SenateB disclosed in the latter a far more serious error in the content than had previously been observed with these materials.

Conclusions

This study used the Levenshtein edit distance (LED) metric to interrogate variants of the 1870 Okmulgee Constitution and it was rewarded in the process by a better understanding of the initial document and its ensuing text reproductions, and by exploring further applications of the LED approach in an interrogation of the 1870 instrument and its related 1875 revision.

Overall, the assortment of Okmulgee Constitution variants suffered from the kinds of spelling errors that interfere with all attempts to publish pristinely, but the repercussions of contaminated text carries more significance in legislative materials than it does in popular press products. In particular, the preamble of the Council version contained the term formes — under the formes prescribed by the treaties of 1866 (Journal of the General Council of the Indian Territory, 1871, p. 44) — that echoed the presence of the term Supintent in the identical document section of the Treaty of Fort Laramie with Sioux, etc., 1851 (Bernholz and Pytlik Zillig, 2009). This kind of syntax error has saturated other international documents, including the original United States Constitution. In that instrument, the state name Pennsylvania was published as Pensylvania; contemporary spellings — chuse vs. choose — were present; and an unnecessary possessive emerged in Article 1, §10: except what may be absolutely necessary for executing it's inspection Laws. Aust remarked on the correction of errors that surface in today's diplomatic documents by suggesting that these were induced by "time pressure" (2005, pp. 110-111). Such problems faced our own Constitutional Convention in 1787, the events leading up to the Constitution of the Confederate States, and perhaps the General Council of the Indian Territory as well, let alone the printers of Okmulgee.

The additional typographical errors of agains, thist, and cammission were manifestations of printing incongruities that have existed even after the development in the nineteenth century of advanced mechanical typesetting options (Huss, 1973). While it is unknown how Excelsior Book and Job Printing might have struggled to set the original Okmulgee Constitution for the Journal of the General Council of the Indian Territory, the invert error of assempled in Beckett emerged sixty years later at Harlow Publishing in Oklahoma City. Comparable imperfections were created in the publishing sphere long before these General Council meetings. The Thou shalt commit adultery blunder in the 1631 King James Bible is prototypic (see Gimcrack, 1833, p. 103), but syntax difficulties also surfaced in works from the fields of astronomy (Talcott and Walker, 1839, p. 249n) and chemistry (Hofmann, 1860, p. 586). In the succeeding century, errors in journal articles describing Virginia state statutes (Editorial, 1912), and in the areas of music (Martino, 1962), mathematics (see the "printer's error" in Douglas [1951]), and history (Jeffrey, 1990) were noted. With regard to specific titles, inquiries into reports of copies of Christopher Columbus' letter announcing his discovery of the New World (Jane, 1930), of the three hundred editions of The Complet Angler (Oliver, 1947), and of the spelling in Milton's works (Shawcross, 1963) demonstrated the broad variety of problematic interpretations. The declaration in the impeachment parameters of Article II, §15 of SenateB — no person shall be convicted with the concurrence of two-thirds of the members present — seems almost insignificant in comparison to the adultery directive contained in that 1631 Bible, but then again, the Okmulgee Constitution was a political document whose concepts were neither ratified nor implemented; these proposed impeachment processes remained as hypothetical constructs only.

Neavill (1975, pp. 29-30) spoke directly to these perils of publishing virtually any manuscript when he remarked that "[k]nowledge is affected at the stage of reproduction by the errors that seem inevitably to creep in whenever a text is reproduced. From the hand copyists of the ancient world to the latest computer composition techniques of today, the reproduction of texts has always involved the introduction of error. It is the responsibility of the publisher (and the printer and author) to eliminate as many of these errors as possible. Conscientious proofreading can greatly reduce the number of errors introduced at this stage, yet almost always some errors remain. Usually they are not as serious as in the so-called Wicked Bible of 1631, in which the 'not' was omitted from the Seventh Commandment. But with a work of any length, the totally error-free text seems to be an ideal which can perhaps be approached but rarely achieved."

It is the "almost always some errors remain" setting that makes text analysis such a critical matter, because the errors individually guide a way through the various renditions to help construct a greater understanding of the primary material itself. While the spirit and the letter of the law was not violated by the exclusions or incursions in the Treaty of Fort Laramie with Sioux, etc., 1851, their presence produced an insight into the thought of the creators and of the reproducers of this instrument. The same may be said now about the Okmulgee Constitution: the inaccuracies supply not only an initial notion of the perceptions and the workmanship of the tribes in Indian Territory in 1871, but also of the purveyors of subsequent popular press exemplars, such as those found in Chronicles and Beckett, that were released to inform greater numbers of subsequent readers.

An editorial comment, concerning Virginia state laws affected by error in 1912, may be a relevant model for consideration. In that statement, the Virginia Law Register observed that "the County of Albemarle has vanished and the County of Albermarle now stands in its place all through the printed Acts of 1912…. If an error so palpable and so inexcusable should be repeated more than once in the volume bearing the stamp of the State's official, may there not be others of greater moment?" (1912, p. 225; emphasis added). A century after this observation, variants still occur, but in many cases, their ultimate effects are still unknown. Hill concluded that "a critical text is an instrument for communicating certain data to a particular audience. That the process of transmission transforms the data [is taken] as axiomatic. So defined, a text will be critical to the degree to which it faithfully transmits those data determined to be of significance to the audience anticipated, making it clear what is transmitted, what suppressed, in full detail, as well as the principles on which this has proceeded. All three terms in this equation — evidence, medium, audience — serve to determine the nature of the final text.... The writing of specifications for an edition and their realization in the texts actually printed are distinct, though related, operations. Specifications are prescriptive and absolute; actual texts rarely are. The best one can hope for is a careful and judicious weighing of the demands of evidence, medium, and audience" (1978, pp. 259-260).

To complement the assertion made at the beginning of this study — that the variant is the lifeblood of text analysis — one must consider Hill's corollary that "[t]he lifeblood of proofreading is the perception of error" (p. 248). This is particularly so for the transmission of legal content, where the demand for textual fidelity is unequivocal, yet the Okmulgee Constitution specimens seen in this study were profound reminders that such materials too suffer from inaccuracies, just as do the works in other genres that permit far more interpretive flexibility. One bright spot in this sea of misprinting was the execution offered by the Choctaw newspaper, The Vindicator, on 21 and 28 June 1873. Table IIIA | Download Excel File reports just 33 errors encompassing 96 total bytes in that presentation, where 23 bytes (24%) of that cumulative LED were due to the incidence of three fourths instead of three-fourths and of for ever rather than forever at line number 2958 and 2959, and 3132 and 3133, respectively. An additional five bytes of benefit were derived from the appropriate use of forms, this, against, commission, and organize, as delivered in the other, non-Council variants. This high degree of fidelity for this instrument — relative to that observed in federal attempts — seems especially fitting for a tribal newspaper published in the wilds of Indian Territory.

Suppleness was evident, however, in the use of the LED approach for comparative testing of the original with the revised Constitution, where the demands to assess identicalness in repetitive 1870 or 1875 companion editions gave way to a need for an effective editorial tool to traverse and align two related yet nevertheless modified documents. Just as the federal Constitution was altered during its evolution during the Constitutional Convention in 1787, so too was the later Okmulgee Constitution by Council representatives, as evidenced by a longer more expressive preamble and by internal amendments. The dissimilar unpadded text lengths — 3,826 vs. 4,060 tokens — announced immediately that an LED test concerning Council and Sixth would herald abundant dissimilarity, but LED scores have now been shown to furnish an index of beneficial change as well.

In the trial for the Council and Sixth accounts, successive small steps of realignment, taken as part of a reasonable approach to coordinating the two forms, produced diminishing cumulative LED values that confirmed convergence. This editorial advantage — in which a quantitative, instead of a qualitative, measurement of textual linkage was the gauge — was not only simple to employ but was intuitive as well. The quest to bring all those Canterbury Tales into register has always been impeded by the disturbing knowledge that the scribes created their descriptions without regard for absolute replication of text order (Spencer, Bordalejo, Wang, Barbrook, Mooney, Robinson, Warnow, and Howe, 2003, pp. 97-98). Today, assessment allows for the resorting of material segments as one judicious way to acquire a final grasp on content. In a recent article, Schmidt and Colomb (2009, p. 498) tracked the controversy of administering representations of such assorted materials that have been placed online. In that endeavor, they re-raised the old question of "What exactly is the text of a work that exists in multiple versions?" and concluded that "[t]he problem of how to represent overlapping hierarchies in markup systems… is simply a subset of the larger problem of how to represent different versions of a work in digital form." They recommended a "model [that] can be visualized… more simply and practically as a list or array of ordered pairs, each consisting of a set of versions and a fragment of text or data... which can be searched, compared, displayed, and edited" (p. 512; emphasis added). This latter activity was realized when the LED tool was used to converge, in a progressively tighter manner, these texts of the Council and the Sixth renditions. Perhaps the future will bring an opportunity for investigators to make simultaneous measurements across these dozen or so renderings of Okmulgee through an online application employing Schmidt and Colomb's scheme. In the interlude, Levenshtein's forty-five year old metric may be a useful tool to serve as a text comparison process that is effective, simple, robust, and intuitive, whether for editorial purposes — such as tuning the linkage between Council and Sixth — or not.

A final word is required to strengthen the perception of the accomplishments of the Okmulgee Constitution creators. The Five Civilized Tribes were removed to Indian Territory before this instrument was developed, yet they had been identified as a unique assembly of American Indians as early as 1775 by James Adair (Williams, 1930). Their sophistication — in virtually all matters — placed them in a special and critical role during interactions between other tribes and the federal government as exemplary models for potential Indian citizens. They developed remarkable social systems in the Southeast long before removal; they created informed treaties with the United States and the Confederate States (and they were always prepared to defend their position in the courts, if necessary); they were employed by both of these governments — in an appropriate acknowledgement of their unique status — to serve as the cajolers of the "wild tribes;" they were directed to create along with other Indian Territory nations the legal groundwork for the development of an Indian state that would be on an equal footing with the rest of the members of the Union; and they responded to that last opportunity with a document that came to serve as the constitutional model for the State of Oklahoma, the state that took their lands and their political seat in the United States. These were not the crazed Indians that Harper's peddled to readers, but a gathering of predominately well-organized communities demanding a sound future for their peoples: fully thirty-two different entities participated in at least one of those Council sessions, a number that was almost one-half of all the Indian groups ever resident in the Indian Territory (Wright, 1951, p. 4).

Aspects of the Okmulgee Constitution served as archetypal proposals for the Sequoyah Constitution, and suggestions from both instruments contributed to Oklahoma's final Constitution. Many ideas were harvested and revitalized for the twentieth century, even though the road to statehood had been a difficult one for all; see Nesbitt (1936) for an informative view of the Sequoyah and the Oklahoma conventions, and Maxwell (1950a and b) for more on the earlier meeting. Fittingly, one need only examine Article VI, §35 of the State's Constitution to learn of the continuing presence in everyday life of the Five Civilized Tribes:

"Description of seal. In the center shall be a five pointed star, with one ray directed upward. The center of the star shall contain the central device of the seal of the Territory of Oklahoma, including the words, 'Labor Omnia Vincit.' The upper left hand ray shall contain the symbol of the ancient seal of the Cherokee Nation, namely: A seven pointed star partially surrounded by a wreath of oak leaves. The ray directed upward shall contain the symbol of the ancient seal of the Chickasaw Nation, namely: An Indian warrior standing upright with bow and shield. The lower left hand ray shall contain the symbol of the ancient seal of the Creek Nation, namely: A sheaf of wheat and a plow. The upper right hand ray shall contain the symbol of the ancient seal of the Choctaw Nation, namely: A tomahawk, bow, and three crossed arrows. The lower right hand ray shall contain the symbol of the ancient seal of the Seminole Nation, namely: A village with houses and a factory beside a lake upon which an Indian is paddling a canoe. Surrounding the central star and grouped between its rays shall be forty-five small stars, divided into five clusters of nine stars each, representing the forty-five states of the Union, to which the forty-sixth is now added. In a circular band surrounding the whole device shall be inscribed, 'GREAT SEAL OF THE STATE OF OKLAHOMA 1907'" (The great seal of the state of Oklahoma, 1957, p. 250; see also Constitution of the state of Oklahoma, 1908, p. 30).

This description was taken almost verbatim from the Sequoyah Constitution of 1905, in which section 1 of Article XVI stated:

"In the center shall be a five-pointed star, with one ray pointing downward. The star shall be divided into five diamond-shaped rays by lines connecting the angles between the rays with the center. The upper left-hand ray shall contain the symbol from the ancient seal of the Cherokee Nation, viz., a seven-pointed star surrounded by a wreath of oak leaves. The upper right-hand ray shall contain the symbol from the ancient seal of the Creek Nation, viz., a sheaf of wheat and a plow. The lower left-hand ray shall contain the symbol from the ancient seal of the Choctaw Nation, viz., a tomahawk, bow, and three crossed arrows. The lower right-hand ray shall contain the symbol from the ancient seal of the Seminole Nation, viz., a village with houses and factory beside a lake upon which an Indian is paddling a canoe. The lowest ray shall contain the symbol from the ancient seal of the Chickasaw Nation, viz. an Indian warrior standing upright with bow in his hand. Surmounting, the star between the two upper rays shall be a half-length figure of Sequoyah holding a tablet upon which are inscribed the letters A J J Q C in the alphabet invented by Sequoyah, and forming the Cherokee words meaning "We are brethren." Surrounding the central star and grouped between its rays, shall be forty-five small stars, representing the forty-five States of the Union to which the forty-sixth is now added. In a circular band surrounding the whole device shall be inscribed 'Great Seal of the State of Sequoyah, 1905'" (Proposed state of Sequoyah, 1906, p. 82).

This latter seal and the other strategies contained in the Sequoyah Constitution were ratified in the Indian Territory by a vote of 56,279 to 9,073 in November 1905. This majority position and the foundation for a strong Indian state were fully presented in a memorial to Congress in 1906 — everything was in place: the eastern segment of what had been Indian Territory was known to be large in area; well populated; and blessed with resources. Further, it was argued, the tribes "constitute[d] a separate and distinct community from any other on earth, with a different history, associations, ideals, and hopes" (Proposed state of Sequoyah, 1906, p. 1), but the political climate had swung towards another path to statehood. The tribes were not forgotten or re-removed. Rather, their thoughts and experiences were accumulated from those constitutional meetings and their pasts were used in part to contribute to an alternative instrument that has given direction to a more traditional state within the United States. Oklahoma certainly embraced its Indian past, but the Indian Territory never became what had been promised to the tribes during the century before.

The words and the concepts of the Okmulgee Constitution, no matter how jumbled or misrepresented in print on occasion, were the sincere effort of these Indian Territory occupants during the last few decades of the nineteenth century. These expressions were of sound integrity and truly mattered, because their contents were to help acquire a true, federally promised Indian domain. The Cherokee, Choctaw, Creek, and Seminole (the Chickasaw did not re-appear after the second annual meeting) carried these proceedings from the beginning to the end. They were aided by other tribes such as the Absent Shawnee, Osage, Ottawa, Peoria, Quapaw, Seneca, and Wyandot whose representatives only missed one or two assemblies and who demonstrated through this unison the true sense of internationalism envisioned by these societies drawn from all parts of the United States. Efficacious internationalism, within a single state, would have been a meaningful exemplar for the United States and its ensuing immigration chaos. Wright had already perceived this when she wrote in 1936 "[t]oday the future of the Oklahoma Indian is in education and in the continued progress of Christian civilization, together with the preservation of the best in native traditions and customs that produced strong leaders and a great art. It is through such forces as these that the Indian has contributed and will continue to contribute to real American culture which will flourish and blossom for ages to come" (p. 161), but with ratification and a chance to be implemented, the Okmulgee Constitution might have contributed more to this prosperity than it did.