On March 16, 1854, the Omaha Indian Tribe entered into a treaty with the United States in which certain lands, including 2,900 acres of land located in the then Territory of Nebraska in an area known as Blackbird Bend, were reserved by the Tribe as part of an agreement which ceded to the United States all other land west of the “centre of the main channel of said Missouri river . . 1 Act of March 16, *6221854, Art. 1,10 Stat. 1043. At the time the Omaha Indian Reservation was established the reserved land within Blackbird Bend was situated on the west side of the Missouri River. However, by 1923 the river had moved more than two miles to the west of the original boundary line so that much of the land contained within the original Blackbird Bend area was • situated on the east side of the river. The defendants assert that early movements of the Missouri River had completely washed away the Reservation lands and that the land now existing within the former boundaries of the Barrett Survey is soil which has accreted to the Iowa riparian land.
As a result of this dispute the United States and the Omaha Indian Tribe, a duly organized corporate body, in 1975 sought equitable relief asserting their right to the 2,900 acres of land now situated in Monona County, Iowa. The United States throughout this litigation acts in the capacity of trustee of the Tribe’s reservation lands.2 The defendants claim title to the land in dispute and seek by way of counterclaims to quiet title in their names. The defendants are Roy Tibbals Wilson, Harold Jackson, a tenant of Roy Tibbals Wilson, Harold Sorenson, Luea Sorenson, Darrell L. Sorenson, Harold M. Sorenson, Charles La-kin, Florence Lakin, the State of Iowa and the State Conservation Commission, R.G.P. Incorporated, Travelers Insurance Company, mortgagee of R.G.P., and Otis Peterson, a tenant of R.G.P.
From at least the 1940’s until April 2, 1975, the defendants and their predecessors in title had occupied, cleared and cultivated the land in dispute. After April 2, 1975, with the assistance of the Bureau of Indian Affairs and with the approval of the United States, the Omaha Indian Tribe seized possession of the land and is presently farming it. After the Tribe had seized the land the United States District Court, the Honorable Edward J. McManus presiding, granted a preliminary injunction permitting occupancy of the land by the Tribe during the pendency of this litigation, but requiring certain accounting procedures pertaining to the crops grown on the land be instituted. Following trial and entry of judgment in favor of the defendants, this court on May 13, 1977, entered a stay pending appeal maintaining in effect the terms of the district court’s preliminary injunction.
After a lengthy trial the district court, the Honorable Andrew W. Bogue presiding, found that the boundary of the Omaha Indian Reservation had shifted with the movements of the Missouri River and quieted title in the defendant landowners. The court found that the plaintiffs had failed to prove that the river movements were controlled by the doctrine of avulsion and held that the river had changed by reason of the erosion of reservation land and accretion to Iowa riparian land. United States v. Wilson, 433 F.Supp. 67 (N.D.Iowa 1977). The district court supplemented its findings on the merits with an opinion resolving choice of law problems, setting forth principles governing avulsion and accretion and discussing the allocation of the burden of proof. United States v. Wilson, 433 F.Supp. 57 (N.D.Iowa 1977).
1. The Issues.
The dispute centers on the ownership of land in an area known as Blackbird Bend *623which was surveyed in 1867 by T. H. Barrett3 on behalf of the General Land Office of the United States.4 The basic issue on appeal is whether the boundary of the reservation remained at its 1854 location despite the significant changes in the location of the Missouri River since that time.
We vacate the judgment of the district court rendered in favor of the defendants; we find the trial court erroneously placed the burden of proof on the Omaha Indian Tribe and failed to properly apply governing principles of federal law relating to avulsion and accretion; we hold the evidence is too speculative and uncertain to show that the reservation boundary shifted by reason of accretion and that the defendants have failed to overcome the presumptive right of possession and title in the Tribe to the reservation lands,
II. The Historical Facts.
The Treaty of 1854 established the eastern boundary of the reserved land at the center of the main channel of the Missouri River.5 Because the exact location of the thalweg6 of the Missouri River could not *624now be established and since the Barrett Survey was conducted only a few years after the reservation was established, the trial court accepted the Barrett Survey as representing the original location of the reservation’s boundary.7 See Plate I.
From 1854 until sometime near 1875 it is generally accepted that the Missouri River moved east until it reached the Iowa easterly high bank.8 The evidence shows the river was approximately 800 feet wide at that time. The first controversial movement of the river in this case occurred after it reached its easterly position against the high bank.9 By 1879 a survey conducted by the Missouri River Commission showed the river, at high flood stage, to be almost 10,000 feet wide covering as much as two-thirds of the Barrett Survey meander lobe. The thalweg had shifted from against the easterly high bank to a point nearly 6,000 feet to the west. See Plate II.
After 1879 the river moved to the northeast until it reached the Iowa northerly high bank10 sometime near 1912. See Plate III. The second dispute between the parties involved the movement of the river away from its northerly high bank location to a location significantly to the south in the period between about 1912 and 1923. See Plate IV. By 1923 almost the entire peninsula described by the Barrett Survey had been cut off by the river’s movement.
“a natural monument demarking the furthest migration northward of the Missouri River from the northern boundary of the 2900 acres as surveyed by Barrett in 1967. .
The course of the Northerly High Bank is described as follows: The Iowa Northerly High Bank runs southeasterly through SE2 of Section 13, T. 84 N., R. 47 W. and continuing through the NE4 of Section 24, T. 84 N., R. 47 W., then easterly through the NW4 and the N2NE4 of Section 19, T. 84 N„ R. 46 W., thence southerly approximately 2,000 feet through the E2NE4 of Section 29, T. 84 N., R. 46 W., terminating at the point of intersection of the Iowa Easterly High Bank. .
*625[[Image here]]
*626[[Image here]]
*627[[Image here]]
[[Image here]]
*628III. Choice of Law.
At trial the Tribe and the government asserted that federal law should control while the defendants contended that Iowa law should have been applied. The district court, however, applied Nebraska law in evaluating the facts of the case. On appeal the Tribe and the government renew their assertion that federal law should be applied in the resolution of this case. We hold that the governing principles of federal law vary significantly with the trial court’s construction of state law and that the court erred in failing to apply that federal law.
A. Interstate Boundaries.
In Oregon ex rel. State Land Board v. Corvallis Sand & Gravel Co., 429 U.S. 363, 97 S.Ct. 582, 50 L.Ed.2d 550 (1977), the Supreme Court reaffirmed the basic rule that the laws of the several states determine the ownership of the banks and shores of waterways. Id. at 378-79, 97 S.Ct. 582. However, the Court recognized an important caveat to this rule:
If a navigable stream is an interstate boundary, this Court, in the exercise of its original jurisdiction over suits between States, has necessarily developed a body of federal common law to determine the effect of a change in the bed of the stream on the boundary.
Id. at 375, 97 S.Ct. at 589.11
As the Supreme Court noted in Arkansas v. Tennessee, 246 U.S. 158, 176, 38 S.Ct. 301, 306, 62 L.Ed. 638 (1918):
[Tjhese dispositions are in each case limited by the interstate boundary, and cannot be permitted to press back the boundary line from where otherwise it should be located.
Cf. St. Louis v. Rutz, 138 U.S. 226, 250, 11 S.Ct. 337, 34 L.Ed. 941 (1891).
Federal common law is applicable even where only a single state is involved in a controversy with a private party, see Cissna v. Tennessee, 246 U.S. 289, 38 S.Ct. 306, 62 L.Ed. 720 (1918), or where only private parties are involved, see Hinderlider v. La Plata River & Cherry Creek Ditch Co., 304 U.S. 92, 58 S.Ct. 803, 82 L.Ed. 1202 (1938); Committee for Consideration of Jones Falls Sewage System v. Train, 539 F.2d 1006, 1009 n. 8 (4th Cir. 1976); Port of Portland v. An Island In Columbia River, 479 F.2d 549 (9th Cir. 1973); Sherrill v. McShan, 356 F.2d 607 (9th Cir. 1966); Iselin v. La Coste, 139 F.2d 887 (5th Cir.), cert. denied, 321 U.S. 790, 64 S.Ct. 791, 88 L.Ed. 1080 (1944), as long as the interests of more than one state are sufficiently implicated in the potential outcome. The rendering of a decision in a private dispute which would “press back” an interstate boundary sufficiently implicates the interests of the states to require the application of federal common law.
In this case any claim that the reservation’s eastern boundary had changed would of necessity have concerned the interstate boundary between Iowa and Nebraska, at least until 1943, thereby invoking federal law since both boundaries were located at the thalweg of the Missouri River. However, in 1943 Iowa and Nebraska entered into a compact under which the boundary between the states was permanently established at the middle of the main channel of the Missouri River in essentially its position in 1943. Iowa-Nebraska Boundary Compact, Iowa Code 1971, p. lxiv; 1943 Iowa Acts ch. 306; 1943 Nebraska Laws ch. 130. Ratified by Congress in Act of July 12, 1943, 57 Stat. 494 (1943).12 To apply the compact it is nevertheless necessary to es*629tablish title good in one state or the other as of 1943. Iowa-Nebraska Boundary Compact § 3. Good title in a state prior to 1943 in turn depends upon the location of the thalweg of the Missouri River, a determination which would have been controlled by federal law. Thus, in this case, since the issue concerns who held good title to the land in question prior to 1943, federal law must be applied.
B. Indian Law.
An equally compelling reason for applying federal law is the special relationship between the United States and the Omaha Indian Tribe and the nature of the interest litigated. The trial court rejected this position under the authority of Fontenelle v. Omaha Tribe of Nebraska, 298 F.Supp. 855 (D.Neb.1969), aff’d, 430 F.2d 143 (8th Cir. 1970), where the Nebraska federal district court applied Nebraska law in an accretion-avulsion dispute between the Omaha Indian Tribe and individual Indians who traced their title back through individual patents issued to their predecessors by the United States.13 Instead, the trial court, finding no federal regulatory program involved14 and no specific act of Congress which displaced state law, reasoned, citing Herron v. Choctaw & Chickasaw Nations, 228 F.2d 830 (10th Cir. 1956), and Francis v. Francis, 203 U.S. 233, 27 S.Ct. 129, 51 L.Ed. 165 (1906), that local law governed title disputes between the Indian tribe and private claimants. 433 F.Supp. at 61.
It has long been held that the rights and incidents of ownership attaching to grants made by the United States of public lands bounded on streams or other bodies of water, navigable or non-navigable, made without reservation or restriction, are to be construed as to their effect according to the law of the state in which the land lies. The fact that a conveyance disposes of tribal lands of Indians under guardianship does not alter the rule. See Oklahoma v. Texas, 258 U.S. 574, 595, 42 S.Ct. 406, 66 L.Ed. 771 (1922). The Fontenelle decision and the other cases cited by the trial court fall within this settled doctrine. In Packer v. Bird, 137 U.S. 661, 669,11 S.Ct. 210, 212, 34 L.Ed. 819 (1891), the Court observed;
The courts of the United States will construe the grants of the general government without reference to the rules of construction adopted by the States for their grants; but whatever incidents or rights attach to the ownership of property conveyed by the government will be determined by the States, subject to the condition that their rules do not impair the efficacy of the grants or the use and enjoyment of the property by the grantee.
The present dispute is not related to incidents or rights flowing from a conveyance of public land or related to a patent grant of Indian allotment lands. Instead, the direct challenge made by the Iowa landowners here affects the boundary line to the reservation land itself, as it was originally contained in the Barrett Survey and established by the Treaty of 1854. The claims asserted by the defendants attempt to extinguish the aboriginal rights of the Omaha Indian Tribe, guaranteed by treaty, in these lands.15 Here the Omaha Indian Tribe *630claims its right to occupy and possess the lands in question arises under federal law. Presumptively, at least, this right has never been extinguished. See discussion of 25 U.S.C. § 194 infra. Under the circumstances the Supreme Court’s observation in Oneida Indian Nation v. County of Oneida, 414 U.S. 661, 677, 94 S.Ct. 772, 782, 39 L.Ed.2d 73 (1974), is applicable here:
In the present case, however, the assertion of a federal controversy does not rest solely on the claim of a right to possession derived from a federal grant of title whose scope will be governed by state law. Rather, it rests on the not insubstantial claim that federal law now protects, and has continuously protected from the time of the formation of the United States, possessory right to tribal lands, wholly apart from the application of state law principles which normally and separately protect a valid right of possession.
State law dealing with riparian rights cannot unilaterally extinguish or deprive Indians of their tribal lands. The land area involved in this appeal relates solely to the original reservation land. Therefore, germane here is the Supreme Court’s statement in Oneida that: “There being no federal statute making the statutory or deci-sional law of the State of New York applicable to the reservations, the controlling law remained federal law; and, absent federal statutory guidance, the governing rule of decision would be fashioned by the federal court in the mode of the common law.” Id. at 674, 94 S.Ct. at 781.16 Riparian ownership rights have been specifically held to be controlled by federal law where trust land is involved.
The nature and extent of riparian rights, if any, in the bed and banks of navigable waters is generally a matter of state law. This is a consequence of the rules that (1) the United States holds title to the bed and banks of navigable waters in trust for future states; and (2) upon admission of a state to the Union, the United States relinquishes to the state the ownership of the bed and banks of its navigable waters. The south half of Flathead Lake presents an exception. Title to the bed and banks of the south half of Flathead Lake below high water mark is held by the United States in trust for the Tribes. Thus, the basis for state determination of riparian rights is nonexistent. State law, therefore, is not applicable.
Confederated Salish & Kootenai Tribes v. Namen, 380 F.Supp. 452, 461 (D.Mont.1974), aff’d, 534 F.2d 1376 (9th Cir.), cert. denied, 429 U.S. 929, 97 S.Ct. 336, 50 L.Ed.2d 300 (1976) (citation omitted).17
*631See also United States v. Finch, 548 F.2d 822, 832-33 (9th Cir. 1976), vacated on other grounds, 433 U.S. 676, 97 S.Ct. 2909, 53 L.Ed.2d 1048 (1977). Cf. Bauman v. Choctaw-Chickasaw Nations, 333 F.2d 785, 787-89 (10th Cir. 1964), cert. denied, 379 U.S. 965, 85 S.Ct. 658, 13 L.Ed.2d 559 (1965).
Finding the land in dispute affects an interstate boundary at the time the controversial movements occurred, and because the Tribe’s right asserted to Indian trust land arises under federal law, we hold that the governing law is federal law.
IV. Burden of Proof.
Section 194 of Title 25 of the United States Code18 provides:
Trial of right of property; burden of proof
In all trials about the right of property in which an Indian may be a party on one side, and a white person on the other, the burden of proof shall rest upon the white person, whenever the Indian shall make out a presumption of title in himself from the fact of previous possession or ownership.
(Emphasis added).
The trial court reasoned that application of the statute to accretion or avulsion cases would be “unreasonable and circuitous” and “inextricably entwined with the merits.” The court’s analysis was apparently based on the idea that to apply § 194 would require the court to presume that the land originally occupied by the Indians within the Barrett Survey is exactly the same land in place today. To do so, the court believed, would be to decide the merits and compel it to decide that the same land remained by reason of avulsive movements of the river. On the other hand, the court reasoned, if the land had washed away and new land had accreted to the Iowa riparian owners, the Indians had never “possessed” the new land and it would not be proper to apply the statute.
We reject this reasoning.
Application of § 194 is not a self-answering inquiry to the issues at hand.19 To hold otherwise, one must presume that the reservation land has in fact been destroyed. Furthermore, the trial court’s reasoning would negate the application of the § 194 statutory burden upon a pleading that simply recites Indian land had been destroyed by the erosive action of a river. Thus, under the trial court’s rationale a party making claim to Indian land could defeat the congressional mandate by mere allegation without proof. We cannot accept the proposition that congressional policy can be so easily thwarted.
It is undisputed that the 1854 treaty established the Tribe as the legal titleholder to the land area within the Barrett Survey lines. This historical fact shows “previous possession or ownership” and is sufficient to raise a presumption of title in the Tribe under the statute and to place the burden of proof on the defendants. Contrary to *632the trial court’s statement, applying the statutory burden of proof in this case does not decide the merits of the case since the fundamental issue still remains: Was there an alteration in the original boundary by reason of the marked movement of the Missouri River in the time periods involved? The defendants must bear the burden of proof that the boundary has been changed.
The early Indian Non-Intercourse Acts provided special treatment to Indian nations as the frontier was being settled. Section 194 was one of many special protective measures included in these Acts. The legislative history of the Act of June 30, 1834, 4 Stat. 729 (1846), of which § 194 was a part, clearly evidences a protectionist policy with regard to Indians.20
The time has been when conciliation was sought; but the time is now passed when the fear of Indian hostility should be a leading feature of our Indian intercourse. Our relation to them is now that of the strong to the weak, and demands at our hands a more liberal policy, as well directed to promote their welfare as our political interests.
H.R.Rep.No.474, 23d Cong., 1st Sess. 10-11 (1834).
The practice of safeguarding Indians in special areas of legislation continues today. See, e. g., DeCoteau v. District County Court, 420 U.S. 425, 444, 95 S.Ct. 1082, 43 L.Ed.2d 300 (1975); Antoine v. Washington, 420 U.S. 194, 199-200, 95 S.Ct. 944, 43 L.Ed.2d 129 (1975).
Defendants urge that notwithstanding the failure of the court to apply § 194, the court, after placing the burden of persuasion on the defendants to establish their counterclaim, found that the defendants had in fact proven by the preponderance of evidence their title in the property.21 Although the trial court observed that it felt no presumption22 aided either party, the *633defendants’ reasoning, as adopted by the trial court, erroneously focuses on the failure of the Tribe to carry a burden of proof to establish title to the land. In discussing the factual evidence, the court’s opinion focuses almost entirely on the inability of the Tribe to prove that the movement of the thalweg was brought about by avulsion.23
We find that the trial court improperly placed the burden of proof on the Indians in the instant case. Title to the Blackbird Bend area as depicted by the Barrett Survey was presumptively shown to be in the Tribe and therefore, notwithstanding the subsequent movement of the thalweg of the Missouri River, the non-Indian claimants were required to assume the burden of proof to show that the Indians no longer had lawful title to the reservation land in question.
We now proceed to an examination of the governing principles of law.
Y. Law of Accretion and Avulsion.
It is fundamental that:
[WJhere running streams are the boundaries between States, the same rule applies as between private proprietors, namely, that when the bed and channel are changed by the natural and gradual processes known as erosion and accretion, the boundary follows the varying course of the stream . .
Arkansas v. Tennessee, 246 U.S. 158,173, 38 S.Ct. 301, 305, 62 L.Ed. 638 (1918).
See also Missouri v. Nebraska, 196 U.S. 23, 34-35, 25 S.Ct. 155, 49 L.Ed. 372 (1904); Nebraska v. Iowa, 143 U.S. at 360-61, 12 S.Ct. 396; Mayor, Aldermen & Inhabitants of New Orleans v. United States, 35 U.S. (10 Pet.) 662, 9 L.Ed. 573 (1836). Equally well settled is the proposition that
[i]f the stream from any cause, natural or artificial, suddenly leaves its old bed and forms a new one, by the process known as an avulsion, the resulting change of channel works no change of boundary, which remains in the middle of the old channel, although no water may be flowing in it, and irrespective of subsequent changes in the new channel.
Arkansas v. Tennessee, supra, 246 U.S. at 173, 38 S.Ct. at 304.
See also Missouri v. Nebraska, supra, 196 U.S. at 35, 25 S.Ct. 155; Nebraska v. Iowa, 143 U.S. at 361, 12 S.Ct. 396.
The trial court found that the eastern boundary of the Omaha Indian Reservation changed with the shifting river since the *634tribe had failed to show that the river had moved by avulsion. In reaching this result the trial court ruled that an avulsion occurs only where a sudden shift in a channel cuts off land “so that after the shift it remains identifiable as land which existed before the change of the channel and which never became a part of the river bed.” 433 F.Supp. at 73. In doing so, the trial court rejected the plaintiffs’ theory that the doctrine of avulsion is equally applicable when a sudden and perceptible shift of the thal-weg occurs within the bed of the stream or over as well as around land in place. The government’s evidence was that such a perceptible shift might occur when the river goes out of its bed and the land is submerged by a flood or freshet. We find the district court too narrowly focuses on identifiable land in place as the sole criterion of avulsion without giving proper weight to the plaintiffs’ theory of their case and to the factual record presented.24
The Supreme Court has defined accretion as “an addition to land coterminous with the water, which is formed so slowly that its progress cannot be perceived . . . Jefferis v. East Omaha Land Co., 134 U.S. 178, 193, 10 S.Ct. 518, 522, 33 L.Ed. 872 (1890).25 In contrast, avulsion has been said to occur in various ways. One observation is that it occurs where there is a “sudden change of the banks of a stream such as occurs when a river forms a new course by going through a bend, the sudden abandonment by a stream of its old channel and the creation of a new one, or a sudden washing from one of its banks of a considerable quantity of land and its deposit on the opposite bank.”26 Ill American Law of Property § 15.26, at 855-56 (1952) (footnotes omitted).
A clear distinction between accretion and the sudden and perceptible movement associated with avulsion is, however, often obscured when applied to the actual movement of uncontrolled rivers. The Supreme Court emphasized the unpredictability of the Missouri River in Nebraska v. Iowa, 406 U.S. 117,119, 92 S.Ct. 1379,1381, 31 L.Ed.2d 733 (1972):
[Ejxperience showed that “the fickle Missouri River . . . refused to be bound by the Supreme Court decree [of 1892]. In the past thirty-five years the river has changed its course so often that it has proved impossible to apply the court decision in all cases, since it is difficult to determine whether the channel of the river has changed by ‘the law of accretion’ or ‘that of avulsion.’ ” Eriksson, Boundaries of Iowa, 25 Iowa J. of Hist, and Pol. 163, 234 (1927).27
*635The early decision in St. Louis v. Rutz, 138 U.S. 226, 11 S.Ct. 337, 34 L.Ed. 941 (1891), illustrates an attempt at more closely defining avulsion in light of actual river conditions. The Court found that violent erosion of shoreland along the Mississippi River between 1865 and 1875 was avulsive in nature, sustaining findings that “the caving in and washing away of the same was rapid and perceptible . . . [occurring] principally at the spring rises or floods of high water in the Mississippi. . . . ” Id. at 231, 11 S.Ct. at 339.28
Rapidity of erosion as the determinative factor in a finding of avulsion was, however, rejected in early dicta in Nebraska v. Iowa, 143 U.S. 359, 12 S.Ct. 396, 36 L.Ed. 186 (1892). See also Oklahoma v. Texas, 260 U.S. 606, 43 S.Ct. 221, 67 L.Ed. 428 (1923) (where the Court applied the rule of accretion, following Nebraska v. Iowa, to the Red River); Philadelphia Co. v. Stimson, 223 U.S. 605, 32 S.Ct. 340, 56 L.Ed. 570 (1912). Although the litigated facts of Nebraska v. Iowa appear to have dealt only with the movement of the Missouri River cutting across the neck of a U-shaped land formation commonly known as an ox-bow,29 the Court expressed its view that the rapidity of the process of subtraction or addition did not prevent application of the rule of accretion.
In discussing the rules of accretion and avulsion the Supreme Court quoted, among others, Vattel, an early civil law authority. Vattel’s formulation of avulsion held that “when the violence of the stream separates a considerable part from one piece of land and joins it to another, but in such manner that it can still be identified, the property of the soil so removed naturally continues vested in its former owner.” 30 143 U.S. at 366, 12 S.Ct. at 398.
*636It is obvious, however, when viewed in the context of the entire opinion, that the Court was not seeking to narrow the traditional scope of avulsion but merely to demonstrate that the rule of accretion, as traditionally defined, was equally applicable to the Missouri River. The Court, in defining avulsion, cited Gould on Waters, noting:
It is equally well settled, that where a stream, which is a boundary, from any cause suddenly abandons its old and seeks a new bed, such change of channel works no change of boundary; and that the boundary remains as it was, in the center of the old channel, although no water may be flowing therein. This sudden and rapid change of channel is termed, in the law, avulsion. In Gould on Waters, sec. 159, it is said: “But if the change is violent and visible, and arises from a known cause, such as a freshet, or a cut through which a new channel is formed, the original thread of the stream continues to mark the limits of the two estates.” 2 Bl.Com. 262; Angelí on Water Courses, § 60; Trustees of Hopkins’ Academy v. Dickinson, 9 Cush. 544; But-tenuth v. St. Louis Bridge Co., 123 Illinois 535 [17 N.E. 439]; Hagan v. Campbell, 8 Porter (Ala.) 9; Murry v. Sermon, 1 Hawks (8 N.C.) 56.
143 U.S. at 361, 12 S.Ct. at 397 (emphasis added).
After Nebraska v. Iowa only a few federal cases have addressed the scope of the avulsion rule in a context other than an ox-bow cut-off involving permanently emerged land in place.31 In Veatch v. White, 23 F.2d 69 (9th Cir. 1927), the facts revealed that:
Years ago, between 1859 and 1874, in the southern shore of Puget Island there was a slough running in a northwesterly direction from the Columbia river. The slough, although only used by fishing boats, had a channel that was shallow and more or less filled with snags. So much of the area of Puget Island as was separated from the mainland by this slough was called Coffee Island, which gradually became submerged. Some time before 1894 the water of the river began to bore out and enlarged the slough, and when freshet waters of 1894 came the slough was so enlarged that a channel formed, which after 1894 was used for navigation. After this new channel was created, shoals formed to some extent on the south, or Oregon, side of Coffee Island, and navigation on that side became unsafe for deep draft ships
Id. at 70 (emphasis added).
Relying on the definition of avulsion in Nebraska v. Iowa, the court held that:
[T]hough by erosion of Puget Island the river has widened and the center of the old channel has been changed somewhat, and has become more shallow than it was at the time of the fixing of the boundary of the state of Oregon, such changes are not to be confused with changes made by the creation of the slough channel, which was caused by sudden and known causes, not by accretion. The demarking line must therefore remain the center of the channel between Puget Island and Oregon before the avulsion.
Id. at 71 (emphasis added).
This court applied the same rationale in Uhlhorn v. U. S. Gypsum Co., 366 F.2d 211 (8th Cir. 1966), cert. denied, 385 U.S. 1026, 87 S.Ct. 753, 17 L.Ed.2d 674 (1967), where the end of a meander bend in the Mississippi River gradually became separated from the mainland area by a small channel. During a flood in 1938 the subsidiary channel was scoured out making it the main navigational channel following the flood. Relying upon Nebraska v. Iowa, this court, through Judges Vogel, Van Oosterhout and Mehaffy, held that, despite the fact that the bar separating the old and new channel was as much as four feet under water when the *637change occurred, the change was avulsive.32 Id. at 219-20. The court observed:
In most instances where a river changes by avultive processes, it has left intervening land above high water mark, but we do not think the elevation of the land mass between an old channel and a new one that is cut by avultive processes is a decisive criterion for a change in a state boundary. By all logic and reason, the boundary should not and does not change from the original thalweg except as the Supreme Court said in State of Arkansas v. State of Tennessee, supra, “by gradual process.” Since there was admittedly nothing gradual here, we conclude and believe that State of Arkansas v. State of Tennessee, supra, commands that the boundary remains in the thalweg of the Bendway Channel subject to its erosion and accretions occurring prior to its stagnation and death.
Id. at 219 (emphasis added).
Several state cases have similarly recognized that the sudden, perceptible change of the channel, whether within or without the river’s original bed, is a critical factor in defining an avulsion.33
Our review of the foregoing authorities leads us to conclude that, although evidence of identifiable land in place may have some probative value that erosion has not occurred, the fact that intervening land may not be visible at the time a sudden flood or freshet occurs is not conclusive in itself.34 To reason otherwise would be to *638limit the rule to the rare situation involving only an obvious neck cut-off where intervening land is not submerged. The history of the rule, the case law developed under it, and the policy underlying the doctrine all support a broader application.35
In Bonelli Cattle Co. v. Arizona, 414 U.S. 313, 327, 94 S.Ct. 517, 38 L.Ed.2d 526 (1973), overruled on other grounds, Oregon ex rel. State Land Board v. Corvallis Sand & Gravel Co., 429 U.S. 363, 97 S.Ct. 582, 50 L.Ed.2d 550 (1977), the Supreme Court explained:
The rationale for the doctrine of avulsion is a need to mitigate the hardship that a shift in title caused by a sudden movement of the river would cause the abutting landowners were the accretion principle to be applied.36
Undisputed historical data relating to the early movements of the Missouri River make clear that the wild and uncontrolled movements of the river did not occur with mathematical precision or follow predictable paths. In fact, as the voluminous testimony and documentary evidence presented by both sides reveal, accretion and avulsion are interrelated phenomena often occurring together and in fact often acting as the motivating force for each other. Erosion and accretion, for example, may change the angle at which a river attacks a downstream bank, increasing the likelihood of an avulsive cut-through. Erosion may narrow the neck of a meander bend producing the necessary conditions for an ox-bow cut-off. Or, as the government asserts, an avulsion can produce river characteristics such as low river current energy areas which are favorable to rapid deposition.
When weighed with the significant policy considerations involved, we hold that, under govering principles, the critical determinant of avulsion is a sudden perceptible shift of the channel.37 Only where the thal-*639weg gradually moves through the intervening land as a direct consequence of erosion and the imperceptible process of accretion to the forming bank do the policies underlying the accretion and avulsion rules justify altering permanent land boundaries to conform with the gradually changing thalweg.
In the present case the plaintiffs claim that a sudden and unusual jump in the thalweg within the bed of a stream or over, as well as around, land (submerged or not) invokes the doctrine of avulsion and its corollary rule that the boundary does not change with the shift of the thalweg. The trial court in rejecting this theory held that a sudden and unusual (erratic) jump or movement of the thalweg without evidence of identifiable land in place falls within the historical rule of accretion. We find this ruling inconsistent with settled principles governing the rule of accretion and the broader parameters involving the doctrine of avulsion. We therefore conclude that it was error for the trial court to reject the plaintiffs’ legal theory in its evaluation of the evidence.
We now turn to the factual findings of the district court.
VI. The 1875-1879 Movement of the Thal-weg.
The basic finding essential to the ' defendants’ case is the trial court’s statement that “the original Omaha Indian reservation land within the 1867 Barrett Survey has subsequently been washed away by the Missouri River . . . and that at the same time new land was added to the Iowa riparian land ... by the gradual process of deposition within the Blackbird Bend area . . . .” 433 F.Supp. at 88. In reviewing this finding our task is not made easier by the district court’s verbatim adoption of the defendants’ analysis of the evidence and proposed findings of fact including the defendants’ credibility assessments of the witnesses.38 We hold the trial court’s conclusion to be clearly erroneous and not supported by substantial evidence.
The defendants’ proof, and the trial court’s conclusion that erosion and accretion produced the movement in the river between 1875 and 1879, rest essentially upon six factual premises. First, defendants assert that bar A (see Plate II) must have been of recent origin (not land in place) since it was situated in the area formerly occupied by the old channel and because willows were the only vegetation growing on the bar. Thus, they contend that the river must have eroded the meander lobe and through the process of accretion deposited the crescent shaped sand bar depicted as bar A during its westward movement. The defendants next point to expert testimony that the remnant channel between bar A and the easterly high bank is a “common phenomenon” associated with erosion and accretion. Such channels are produced, they assert, when a river gradually moves away from its previous channel by the process of erosion and accretion. The accretive deposition then closes the old channel at its upper end forming the remnant channel. Third, defendants’ experts assert that bars B and D shown on the 1879 map could not *640have been identifiable land in place since they were barren of vegetation. Instead, they suggest that the bars were middle sand bars formed behind the southerly and westerly migration of the river. Defendants further contend that since no cottonwood trees, which grow only on land that is stabilized and not subject to frequent inundation, were shown to be growing on bar C, the bar must have been completely eroded away as the river moved. Defendants also find support for their accretion theory in the opinion of one of their experts that the parallel positioning of bars A, B, C and D was consistent with a gradual southwesterly movement of the river as a result of erosion of the meander lobe. Finally, the defendants introduced the experiments of Captain J. F. Friedkin39 into evidence which, along with hydrologic experience, allegedly showed “that during high water flows, the river will flow and erode against the upper side of a point bar . . .” 40
The court concluded on the basis of this evidence that the meander lobe was eroded away as the channel moved over the area where the lobe had existed. Thus, the court ruled that the shift in the course of the river was a consequence of progressive scour and deposition, that is by accretion. 433 F.Supp. at 77-78.
Our analysis of the record leads us to conclude that the evidence presented at trial as a matter of law was insufficient to satisfy the defendants’ burden of proving by the preponderance of the evidence that erosion and accretion were responsible for the significant shifts of the river between 1875 and 1879.
A. Identifiable Land in Place.
The years from 1875 through 1879 were years of extremely high water flow on the Missouri River; no other four year period equaled or exceeded the maximum discharge for those years.41 Defendants’ experts agreed that during this time the meander lobe was low, at times entirely under the surface of the moving river and extremely vulnerable. This is corroborated by Barrett, who in his 1867 survey described the eastern end of the meander lobe as a low sandy point.
Recognizing that identifiable land in place may in given circumstances have probative value in helping to distinguish between accretion and avulsion, that is, at least in the sense of determining whether intervening land has been completely eroded or not, little significance can be attached to its alleged absence under the evidence adduced in the present case. The defendants urge that absence of identifiable land in place provides the necessary inference of erosion of the original reservation land. However, the trial court’s finding that in 1879 bar A was of recent origin and was not identifiable land in place does not afford a permissible inference that the original reservation land had been eroded since, as the trial court itself pointed out, at least part of bar A was located within the area of the old abandoned channel. No land could be eroded if it never existed in the first place. Under the circumstances it is obvious that bar A is nothing more than land formed by deposition after the channel was abandoned.
The evidence that bar D and part of bar B, lying to the west of bar A, and inside the *641Barrett Survey lines, were not identifiable land in place is equally inconclusive. It is conceded that the eastern end of the Barrett Survey was made up of sandy material. Thus, even if the river changed by avulsive movements through the end of the meander lobe, the land remaining in place would have been low and composed of sand.42
The trial court also found that the land which had previously occupied the area shown as bar C on the 1879 map had been completely eroded away and that bar C had formed thereafter as a middle bar as the thalweg moved to the west. The court observed: “If bar ‘C’ were land-in-place which had existed prior to 1879, it would have supported the growth of cottonwoods or other vegetation more substantial than willows by 1890.”43 433 F.Supp. at 77. Although it is possible that the land represented by bar C may have completely eroded, it is entirely speculative to say that that is what occurred. The record also supports the possibility that bar C, located on the eastern end of the lobe, was the same surface area described by Barrett in his notes and was not built up by accretive deposits. The record is insufficient to prove what actually occurred.
Under defendants’ accretion theory bar C would of necessity have been comprised of relatively newly deposited soil since it was located close to the 1879 position of the thalweg. Nevertheless, the record shows that willows were growing on bar C in 1879 indicating, as Dr. McQuivey, a government witness, theorized, that the bar may well have been land in place.44 The fact that cottonwoods were not also growing on the bar does not prove the contrary. Bar C, along with bars B and D, was located near the eastern end of the Barrett Survey which Barrett described as sandy soil and frequently inundated. Testimony in the record shows that cottonwoods did not thrive in areas subject to frequent inundation. We find the evidence concerning bar C to be highly conjectural and inconclusive as to whether it formed by accretion or was in fact identifiable land in place.
There exists another basic reason why we regard the evidence of the defendants as insubstantial. The opinion of the defendants’ experts45 that no identifiable *642land remained in place after the movement of the river sometime between 1875 and 1879 is essentially based upon inferences drawn from the 1879 map admittedly prepared in a 10-day period during the June rise of the Missouri River when the river was as much as five feet above its ordinary high water level. At the time the river was shown to be nearly 10,000 feet wide, whereas in 1875 the bed had been only approximately 800 feet across. Soundings taken at that time demonstrate that a substantial land area was immediately below the surface of the flood water. The defendants do not contend that the river bed permanently expanded to 10,000 feet as shown in 1879; it obviously would be much narrower upon subsidence. See, e. g., the 1890 Missouri River Commission Map, set out as Plate V, infra, in which the river is shown to be no more than 3,500 feet wide. The existence or nonexistence of identifiable land in place could not have been accurately assessed at a time when the river’s flow was abnormally high during floods which completely inundated the adjacent land. Thus, any inferences drawn from the alleged land forms exhibited on the 1879 map appear to be highly conjectural. Substantial evidence cannot be based upon an inference drawn from facts which are uncertain or speculative and which raise only a conjecture or possibility.46 See Polk v. Ford Motor Co., 529 F.2d 259, 271 (8th Cir.), cert. denied, 426 U.S. 907, 96 S.Ct. 2229, 48 L.Ed.2d 832 (1976); Wilson v. Volkswagen of America, Inc., 561 F.2d 494, 517 n. 65 (4th Cir. 1977); Padgett v. Buxton-Smith Mercantile Co., 262 F.2d 39, 41 (10th Cir. 1958); Gilbert v. Gulf Oil Corp., 175 F.2d 705, 709 (4th Cir. 1949).47 As this court indicated in Uhlhorn v. U. S. Gypsum Co., 366 F.2d at 219-20, if land over which a channel changes during abnormal high water periods, inundating all intervening land masses, is identifiable as the same land mass upon subsidence of the high water, the boundary does not change even though the land’s surface may be somewhat eroded.
Defendants’ experts also relied upon inferences drawn from the existence of remnant channel formations, from the parallel positioning of bars A, B, C and D and their interpretation of applicable principles of hydrology. The record, in our judgment, requires us to give little or no probative effect to the ultimate conclusion reached from these factual premises.
B. Remnant Channels.
The defendants rely on evidence of remnant channels in the Blackbird Bend area as a “common phenomenon” associated with the progressive and gradual movement of a channel as a result of accretion. However, this evidence was proffered only as a possible alternative explanation to the plaintiffs’ evidénce that the remnant channel areas *643demonstrated avulsive changes in the thal-weg.48
The evidence further reflects that Dr. John F. Kennedy testified that, as the thal-weg shifted to the west, the low energy level of the river in the former channel created by the movement would result in deposition there. The defendants conclude from this that the shift in the river was “a consequence of progressive scour and deposition.” The testimony of defendants’ expert, Dr. Kennedy, showed, however, that the shift of the thalweg is often the responsible agent for the “subsequent diminished sediment transport capacity of the water.” Thus, the testimony is inconclusive as to whether the remnant channel was formed by accretion or avulsion since a remnant chute might also have been formed after the thalweg suddenly and perceptibly moved. In the latter case the deposition forming the remnant channel would be the effect of the low river energy in the area brought about by the sudden avulsive shift of the thalweg rather than a consequence of accretion. As the Supreme Court observed in Arkansas v. Tennessee, 246 U.S. at 175, 38 S.Ct. at 305,
if the stream leaves its former bed and establishes a new one as the result of an avulsion, the boundary remains in the middle of the former channel. An avulsion has this effect, whether it results in the drying up of the old channel or not. So long as that channel remains a running stream, the boundary marked by it is still subject to be changed by erosion and accretion; but when the water becomes stagnant, the effect of these processes is at an end; the boundary then becomes fixed in the middle of the channel as we have defined it, and the gradual filling up of the bed that ensues is not to be treated as an accretion to the shores but as an ultimate effect of the avulsion.
(Emphasis added).
None of the explanations for the remnant channels are, however, more than sheer conjecture and do not, under the factual circumstance shown here, constitute probative evidence of whether the movement occurred by either accretion or avulsion.49
C. Bar Formation.
Just as the remnant channels may have formed as the effect of an avulsion rather than as the result of accretion, the parallel positioning of bars A, B, C and D, while perhaps characteristic of accretion, could have resulted from the accretion which followed as an effect of avulsion as well. Moreover, as we have discussed the actual surface positioning of those bars based on the 1879 map is conjectural since the contours were drawn at the high water stage which existed at that time.
D. Captain Friedkin’s Experiments.
The final proofs which defendants and the trial court relied on to establish accretion as the cause of the river’s movement were Captain Friedkin’s experiments and general principles of hydrology. Both Friedkin’s experiments and principles of hydrology, however, provide strong evidence that avulsive change may have produced the 1879 movement. Friedkin noted that when a meander bend reaches its limiting width,50 as it is conceded Blackbird Bend *644had here, it does not stop caving its banks. Instead, “[t]he flow short-cuts over the convex bars of bends, chute channels form, and a new bend develops a little farther downstream.” J. Friedkin, Meandering of Alluvial Rivers 14 (1945) (footnote omitted). Such chute channels51 are avulsive since the channel moves around or over, not imperceptibly through the interjacent land. One of the defendants’ witnesses, Dr. George R. Hallberg, testified that it was quite possible that chutes similar to those described by Friedkin could have formed along the eastern edge of Blackbird Bend, although he felt the area was probably submerged when this occurred. Hallberg’s testimony concurred with the government’s expert, Dr. Paul McQuivey, a research hydrologist, who testified that, when a river is at high flow it tends to increase its radius, that is, it tends to take a more direct course rather than a meandering one.52 To do so, he continued, the thalweg moves to the inside of the bend, and erodes new chute channels 53 through the point bar. Eventually, the record shows, one of the chute channels becomes large enough to carry the main flow of the river, and the river abandons the old channel.54 Cf. Veatch v. White, 23 F.2d at 70.
In addition to the evidence suggesting that the movement of the river in 1879 could have occurred through a series of chute channels, the evidence also supports the possibility that during this same time a cut-off could have occurred across the neck of Blackbird Bend.55 Friedkin also described the conditions when a cut-off is likely:
Natural cut-offs occur when a meandering river develops and finally erodes through a narrow neck between the upper and lower arms of a bend. . In over 50 laboratory streams in uniform materials not a single cut-off of this type developed. This fact indicates that natural cut-offs result from local differences in the erodibility of bank materials.
With all bends in a meandering river migrating down the valley at the same rate, a cut-off cannot possibly develop. The upper arm of a bend never catches up with the lower arm. For a natural *645cut-off to develop, erosion on the downstream bank of the lower arm of a bend must be slower than along the upper arm. Generally, when a cut-off develops, erosion in the bend proper has taken place so that the flow in the lower arm of the bend has become directed up-valley . and a narrow neck has developed. It is, therefore, indicated that it is the local differences in erodibility of the bank materials in natural rivers which cause narrow necks to form and cut-offs to develop.
Friedkin, supra at 16.56
There exists strong corroborative proof of such occurrences from Barrett’s field notes made during his 1867 survey. He wrote:
Until very recently, appearances indicated that this point was increasing in size from the deposits and drift of the river; but, during the present season, the river, rising to a great height, partly worked a channel across it, which may, eventually, entirely detach it from the Nebraska shore, rendering it an island in the river.
Tribe exhibit 26E, at 6.
The evidence also demonstrated that the Iowa south bank of the Blackbird Bend area was composed of erosion resistant material which would have prevented the southerly movement of the meander bend and made a cut-off possible. Cf. Friedkin, supra at 16-17.57 In addition, the record shows that the potential for breakthroughs in the 1875 through 1879 period was enhanced by high water flows during those years.58
*646Based on the foregoing analysis we find the factual predicate, supporting defendants’ theory as the causative reason for the river’s movement, is in large part conjectural and the opinions drawn therefrom must be viewed as speculative. Under the circumstances we hold that the defendants have failed to meet their burden of proof that the significant arid marked changes of the Missouri River between 1875 and 1879 were the result of erosion and gradual imperceptible accretion to the Iowa bank.
VII. The River Between 1879 and 1912.
In 1881 the highest flood waters of any' time since reliable records have been kept occurred on the Missouri River. However, no maps were introduced to show the location of the river between 1879 and 1890. The next map available after 1879 is the 1890 Missouri River Commission map, see Plate V, which shows the thalweg to have moved north and somewhat to the east of its 1879 location. Significantly, the river is shown to occupy far less area than that shown on the 1879 map and the land to the east of the thalweg is shown as stabilized soil with permanent vegetation growing on it.
The trial court felt these facts “simply [show] the normal and logical progression of the accretion which was shown to be forming in 1879 . . . .” 433 F.Supp. at 79. There exists no factual predicate whatsoever to support this conclusion.59
After 1890 the river moved in a northerly and easterly direction to a position shown as the northerly high bank in approximately 1912. See Plate III.
We now turn to the trial court’s findings and the factual discussion involving the marked change in the course of the river between 1912 and 1923.
*647[[Image here]]
VIII. The 1912-1923 Movement of the Thalweg.
The trial court found, again relying upon the defendants’ analysis of the evidence, that the 1912 through 1923 movement of the Missouri River was also a direct result of erosion and imperceptible accretion to the Iowa riparian land. The court concluded:
[T]he Court finds that there were no avulsions in the Blackbird Bend area between 1890 and 1923, and that the move*648ments of the river during that period of time were erosive in nature so that accretion was being formed on the side of the river opposite the erosion. The Court finds that after 1890 the river moved erosively until it reached what is now the northerly high bank following which it commenced a southern migration throughout the Blackbird Bend area until it reached its 1923 position. This migration eroded almost all of the westerly end of the land as surveyed by Barrett in 1867 and the deposition which occurred during the southerly migration of the river was accretion to the northerly and easterly high banks and thereby became accretion to the Iowa riparian owners.
433 F.Supp. at 85.
The defendants contend that the river moved in a slow progressive movement south and westward to bring about the remarkable change in the thalweg between 1912 and 1923. Compare Plate III with Plate IV. The basic disagreement between the experts for both sides still exists. The defendants’ experts — Hallberg, Kennedy, and Raymond L. Huber — testified that sometime around 1912 the river moved away from the northerly high bank and began to erode to the south depositing accretion materials to Iowa riparian land. It was their theory that all of the land in the path of the river was eroded away. On the other hand, the plaintiffs’ experts varied in their appraisal of the river movement during this period. Elmer M. Clark, an aerial surveyor, believed that the river achieved its 1923 position by one or more sudden and noticeable movements of the channel. Plaintiffs’ witness, Dr. Charlés S. Robinson, a geologist, thought the southward movement occurred suddenly during one high water period in a matter of a few days or months. Dr. McQuivey, the government’s expert, felt that the river moved as a result of several avulsive point bar cut-offs.
In reviewing the record, we cannot but echo the words of witness Huber, that it is again only an “educated guess” as to just how the river moved. The extremely speculative assessment of the reasons for the river’s movement offered by the defendants is not sufficient to sustain their burden of proof.
The evidence relied upon by the defendants, and the court in reaching its conclusion that accretion caused the 1912-1923 river movement, falls essentially within four areas of discussion. First, the defendants’ experts point to evidence of the absence of identifiable land in place. This evidence consisted of testimony from witnesses who had lived in the Blackbird Bend area at the time of the movement that the land on the Iowa side was “new accretion land.” Defendants’ witnesses also pointed out that there were no geologic or geo-morphic formations evident that were typical of avulsive river movement. In addition, the defendants alleged that no trees could be found which predated the movement away from the northerly high bank.60
Second, defendants’ experts asserted that a change in the Missouri River above Blackbird Bend altered the angle of approach of the river into Blackbird Bend ultimately producing a southerly migration of the river. In support of this point the defendants introduced in evidence a 1912 survey showing a developing sand bar on the Iowa shore (referred to as the Fairchild sand bar) said to be consistent with a changed angle of entry. The defendants’ witnesses also contended that hydrological principles would predict erosive changes in the river as a result of the change of the angle at which the river entered Blackbird Bend. Complimenting this proof were several Indian allotment letters which documented erosion in the northwestern part of the Blackbird *649Bend meander lobe at a point opposite the formation of the Fairchild sand bar.61
Third, topographic cross-sections prepared by Dr. Hallberg showed that the land surface gradient in the Blackbird Bend area sloped from northwest to southeast, and that the land formations in the area were progressively lower in elevation in the southeast. In his opinion, this gradient was compatible with a theory of bar growth on the northeast side of the river as it migrated to the southwest.
Finally, the defendants relied on soil analysis which allegedly revealed findings consistent with the southward and eastward accretive movement of the river.62 The government’s expert, Dr. McQuivey, had testified that analysis of soil samples taken in the area produced data indicating that the soil becomes sandier as one moves westward in the Blackbird Bend area. Dr. Hallberg utilized this data to theorize that the sandier nature of the western edge of Blackbird Bend indicated that remnant channels had formed behind accretive point bars deposited as the river slowly moved to the southwest.63
Very few facts* relating to the actual river movement in the period between 1912 and 1923 are proven on the record. First, it is clear that the thalweg had moved substantially in a relatively short period of time; perhaps as few as three years,64 but no more than 11 years. Second, as the trial court found, substantial high water periods occurred in 1905, 1906, 1912, 1913, 1915, 1916 and 1920. 433 F.Supp. at 83. Third, the angle of approach of the river had changed subjecting the adjacent land in place to at least surface erosion. Furthermore, substantial erosion was undeniably occurring along the Nebraska shore.65 Finally, it is beyond dispute that remnant channel-like formations may be identified by soil samples and aerial photographs. These established facts do not prove that either accretion or avulsion caused the river’s movement; the issue remains whether they may form a sufficient factual predicate for defendants’ witnesses to conclude that the marked change of the thalweg was a gradual one directly caused by erosion and imperceptible deposition to the Iowa land. We hold the evidence too conjectural and the ultimate conclusion reached too speculative to sustain the defendants’ burden of proof under § 194.
The expert opinions and the lay testimony indicating that land in place could not be identified in the area are not sufficient to prove that accretion was responsible for the 1912-1923 movement. It is well established on the record that after 1890 as the river moved in a northerly direction alluvion soil subject to frequent inundation was deposited throughout much of the area. Thereafter, when the river moved southwesterly away from the northerly high bank, either by accretion or as the result of avulsion, this sandy area would not reveal any conspicuous identifiable features. The fact that some erosive action occurred in the *650northwest córner of the Barrett Survey area does not support the inference that accretion was the primary cause for the movement. As we have noted earlier, erosion may readily occur along with avulsive movements of a river; the two phenomena may often take place together. Under the circumstances the absence of identifiable land in place has little probative value in determining whether erosion and accretion were in fact the responsible elements for the 1912 through 1923 movement of the river.66
Aside from the inconclusive evidence relating to identifiable land in place, little can be found in the record that supports the defendants’ position. The fact that the river changed its angle of approach does not in itself support an inference that accretion was responsible for the changed movement of the thalweg. Although the change may have altered the erosion patterns in the Blackbird Bend area, it is still a tenuous conclusion at best to infer from this that the thalweg moved imperceptibly. While reasoned deductions may be made from probative facts, the ultimate conclusion may not rest on mere guesswork. Similarly, as in the 1875-79 river movement, the sand bars and the gradient of the land could have just as easily occurred from deposition resulting from low river energy caused by avulsive movements of the river as by a gradual shift or movement of the thalweg.
The little solid scientific evidence in the record contradicts the defendants’ theory of how the river moved. Soil samples taken in the area by government experts67 corroborate Dr. McQuivey’s testimony of the existence of four major channel remnants.68 Aerial photographs made in 1925 also indicated the channel remnants in the area. Although the defendants presented an alternative theory of how the remnant channels might have formed, their existence is strongly probative of the government’s theory that the river migrated from the northerly high bank to its 1923 position by a series of four point bar cut-offs.
We conclude on the basis of an overall review of the record that it is entirely speculative to determine when or how the thal-weg moved to the position shown on the 1923 map.
IX. Conclusion.
The district court’s ruling, adopting the defendants’ theory of their case, depends on the assumption that the doctrine of avulsion is not applicable without evidence of identifiable land in place. On the record presented we deem this an erroneous legal assumption. When considered in the context of the broader parameters of avulsion we hold the defendants’ case establishes only speculative inferences as to whether the thalweg moved by accretion or avulsion in the critical time periods involved. The essential inferences cannot be left to speculation or *651conjecture. Under the circumstances, we hold that the defendants have failed in sustaining their burden of proof under § 194.69
We recognize that to require the defendants to prove the cause of the river’s movement occurring some 100 years after the event is indeed an onerous burden. This may seem to be an injustice when one considers that the defendants or their predecessors have possessed and continuously farmed the land without protest for nearly 40 years. However, as the trial judge observed in a letter written to counsel after trial, divesting the Omaha Indian Tribe of their original reservation land promotes an equal injustice.
While burdening either side in this case with proving the movements of the river occurring many years ago may result in undesirable hardships, the clear policy of the federal government mandates that the interests of the Omaha Indian Tribe be given their historical and statutory protection. These important possessory land interests cannot be taken away on proof that is basically speculative and conjectural.
The judgment of the district court is ordered vacated; the cause is remanded with directions to enter judgment quieting title in the trust lands involved in this action70 in the United States as trustee, and the Omaha Indian Tribe; the prior orders relating to escrow funds and accounting procedures governing the 2,900 acres within the original boundary of the Barrett Survey as originally ceded to the Tribe in the Treaty of 1854 are ordered dissolved.71
. The treaty was entered into by George W. Manypenny, Commissioner on the part of the United States, ratified by the United States Senate and signed by President Franklin Pierce on the 21st day of June 1854. “Marks” indicating agreement of the Tribe were made by the then Chiefs of the Omaha Indian Tribe: Shon-ga-ska, or Logan Fontenelle; E-sta-mah-za, or Joseph Le Flesche; Gra-tah-mah-je, or Standing Hawk; Gah-he-ga-gin-gah, or Little Chief; Tah-wah-gah-ha, or Village Maker; Wah-no-ke-ga, or Noise; and So-da-nah-ze, or Yellow *622Smoke. In addition to providing for the cession of land to the United States and defining the limits of the Reservation’s boundary, the Omaha Tribe agreed to vacate all other lands acknowledging their complete dependence on the government of the United States. Art. 10. The United States agreed to pay certain monies over the ensuing 40 years and to aid the Tribe by various affirmative means to “advance them in civilization.” Art. 4. The other articles of the treaty include several specific covenants and pledges exchanged between the United States and the Tribe. See Act of March 16, 1854, 10 Stat. 1043.
. While Indians have the right of use and occupancy to tribal lands, the United States holds the land as a trustee for the benefit of the Indians. See Morrison v. Work, 266 U.S. 481, 485, 45 S.Ct. 149, 69 L.Ed. 394 (1925). See also Choate v. Trapp, 224 U.S. 665, 678, 32 S.Ct. 565, 56 L.Ed. 941 (1912); United States v. Ric-kert, 188 U.S. 432, 442-43, 23 S.Ct. 478, 47 L.Ed. 532 (1903). F. Cohen, Handbook of Federal Indian Law 94-95 (AMS Press ed. 1972).
. Barrett’s survey established the meander line for the Nebraska shore of the Missouri River.
. Claims to land outside of an area described by the 1867 Barrett Survey were severed from the present case. The trial court explained the severance as follows:
Approximately 8000 acres of land claimed by the Omaha Indian Tribe in C75-4067 and all issues of damages were severed. The severance of the Barrett Survey Area from the other claims of the Omaha Indian Tribe was necessary because the Barrett Survey line is the only clearly ascertainable line of demar-kation, and was the boundary of the area of which the Tribe received possession by virtue of a preliminary injunction entered June 5, 1975. Thus as to the Barrett Survey Area the action was construed as an equitable quiet title action, and the demands of various defendants'for a jury trial were denied as to that area. As to lands claimed by the Tribe outside the Barrett Survey Area, the action was treated as a legal action for ejectment, in which defendant’s demands for a jury trial may be sustainable. This left the 2900 acres within the Barrett Survey as the subject matter of this trial since the dispute over that land is common to all three lawsuits.
United States v. Wilson, 433 F.Supp. 67, 69 (N.D.Iowa 1977).
. Article One of the treaty provided in part:
The Omaha Indians cede to .the United States all their lands west of the Missouri river, and south of a line drawn due west from a point in the centre of the main channel of said Missouri river due east of where the Ayoway river disembogues out of the bluffs, to the western boundary of the Omaha country, and forever relinquish all right and title to the country south of said line: Provided, however, that if the country north of said due west line, which is reserved by the Omahas for their future home, should not on exploration prove to be a satisfactory and suitable location for said Indians, the President may, with the consent of said Indians,
set apart and assign to them, within or outside of the ceded country, a residence suited for and acceptable to them.
Act of March 16, 1854, Art. 1, 10 Stat. 1043.
. The word thalweg is derived from the German language and is said to be
[t]he channel continuously used for navigation. In other words, the thalweg is not a boundary line but in fact a boundary area, because the channel of a river is never a precise line. De la Pradelle emphasizes this special character in one of his definitions: the thalweg is that specific area in the river which is in practice the variable route followed by boatmen on their way down the river.
Bouchez, The Fixing of Boundaries in International Boundary Fivers, 12 Int’l & Comp.L.Q. 789, 793 (1963) (footnote omitted).
The United States Supreme Court adopted the thalweg principle as the standard rule for establishing interstate boundaries in Iowa v. Illinois, 147 U.S. 1, 10, 13 S.Ct. 239, 37 L.Ed. 55 (1893). Later, in Minnesota v. Wisconsin, 252 U.S. 273, 40 S.Ct. 313, 64 L.Ed. 558 (1920), the Court explained the purpose of the rule.
The doctrine of Thalweg, a modification of the more ancient principle which required equal division of territory, was adopted in order to preserve to each State equality of right in the beneficial use of the stream as a means of communication. Accordingly, the middle of the principal channel of navigation is commonly accepted as the boundary. Equality in the beneficial use often would be defeated, rather than promoted, by fixing the boundary on a given line merely because it connects points of greatest depth. Deepest water and the principal navigable channel are not necessarily the same. The rule has direct reference to actual or probable use in the ordinary course, and common experience shows that vessels do not follow a narrow crooked channel close to shore, however deep, when they can proceed on a safer and more direct one with sufficient water.
*624Id. at 282, 40 S.Ct. at 319.
See also New Jersey v. Delaware, 291 U.S. 361, 381, 54 S.Ct. 407, 78 L.Ed. 847 (1934); Louisiana v. Mississippi, 202 U.S. 1, 49, 26 S.Ct. 408, 50 L.Ed. 913 (1906); Uhlhorn v. U.S. Gypsum Co., 366 F.2d 211, 215 (8th Cir. 1966); Whiteside v. Norton, 205 F. 5, 9 (8th Cir. 1913); 1 C. Hyde, International Law § 138 (2d rev. ed. 1951); 8 Op. Att’y Gen. 175 (1856).
. Although the exact words used in acts of Congress defining the boundaries of a state may vary (for example, “middle of the river,” “middle of the main channel,” “mid-channel” or “middle thread of the channel”) the Supreme Court has presumed that since “[i]t is the free navigation of the river . . . that States demand shall be secured to them”, Iowa v. Illinois, supra, 147 U.S. at 13, 13 S.Ct. at 243, quoting Buttenuth v. St. Louis Bridge Co., 123 Ill. 535, 17 N.E. 439 (1888), in the absence of a contrary agreement, the middle of the main channel, the thalweg, establishes the interstate boundary. Iowa v. Illinois, supra, 147 U.S. at 13, 13 S.Ct. 239.
. There is no disagreement as to the Tribe’s description that:
The Easterly High Bank is a natural monument demarking the furthest point of progression of the eastern migration of the Missouri River from the eastern boundary of the 2900 acres, as surveyed by Barrett in 1867.
It commences at a point on the Easterly High Bank located approximately 2,200 feet southwesterly from the corner common to
Sections 20, 21, 28 and 29, T. 84 N., R. 46 W. of the 5th P.M., continuing down said Easterly High Bank a distance of approximately 3'A miles to a point on the Easterly High Bank approximately 4,900 feet northwesterly of the common corner of Sections 4, 5, 8, and 9, T. 83 N., R. 46 W. of the 5th P.M.
. The exact time at which the river reached and then left the easterly high bank is not certain. All parties agree the 1875 atlas map, see generally Plate I, showing the river against the high bank could have been prepared sometime before 1875.
. The Tribe describes the northerly high bank as
. See also Illinois v. City of Milwaukee, 406 U.S. 91, 105-06, 92 S.Ct. 1385, 31 L.Ed.2d 712 (1972); Arkansas v. Texas, 346 U.S. 368, 372-73, 74 S.Ct. 109, 98 L.Ed. 80 (1953) (Jackson, J„ dissenting); Hinderlider v. La Plata River & Cherry Creek Ditch Co., 304 U.S. 92, 110, 58 S.Ct. 803, 82 L.Ed. 1202 (1938); Note, The Federal Common Law, 82 Harv.L.Rev. 1512, 1520 (1969).
. The compact provided for the cession of land previously lying within the boundaries of one state to the state within which it was located following the establishment of the permanent boundary. Titles, mortgages, and other liens good in the ceding state must be recognized as valid in the receiving state. Iowa Code 1971, p. lxiv; 1943 Iowa Acts ch. 306, *629§§ 2-3; 1943 Nebraska Laws ch. 130, §§ 2-3. See also Nebraska v. Iowa, 406 U.S. 117, 122, 92 S.Ct. 1379, 31 L.Ed.2d 733 (1972).
. Although this court affirmed the district court, the issue of state vis-a-vis federal law was not discussed. This court cited only federal authorities relating to the accretion-avulsion issue.
. See generally United States v. Little Lake Misere Land Co., 412 U.S. 580, 93 S.Ct. 2389, 37 L.Ed.2d 187 (1973).
. In the early case of Worcester v. Georgia, 31 U.S. (6 Pet.) 515, 556-57, 8 L.Ed. 483 (1832), Chief Justice Marshall observed:
From the commencement of our government, congress has passed acts to regulate trade and intercourse with the Indians, which treat them as nations, respect their rights, and manifest a firm purpose to afford that protection which treaties stipulate. All these acts . . . manifestly consider the several Indian nations as distinct political communities, having territorial boundaries, within which their authority is exclusive, and having a right to all the lands within those bounda-
*630ríes, which is not only acknowledged, but guarantied by the United States.
. The special concern for preserving Indian land in the Indians is evidenced in the immunity of trust lands from the traditional restrictions on recovering land such as statutes of limitation, laches, and adverse possession. As the Fourth Circuit Court of Appeals early observed:
The determinative fact is that the. federal government has assumed towards them [the Eastern Band of Cherokee Indians] the same sort of guardianship that it exercises over other tribes of Indians, from which it results that their property becomes an instrumentality of that government for the accomplishment of a proper governmental purpose and may not be taken from them by contract, adverse possession, or otherwise, without its consent. Indeed, a statute of the United States expressly forbids the acquisition of lands of any Indian tribe by purchase, grant, lease or other conveyance, except by treaty
or convention and subjects to penalty anyone not being employed under the authority of the United States who attempts to negotiate such treaty. R.S. § 2116, 25 U.S.C.A. § 177.
United States v. 7,405.3 Acres of Land, 97 F.2d 417, 422 (4th Cir. 1938) (citations omitted). See also United States v. Candelaria, 271 U.S. 432, 440 — 42, 46 S.Ct. 561, 70 L.Ed. 1023 (1926); United States v. Minnesota, 270 U.S. 181, 196, 46 S.Ct. 298, 70 L.Ed. 539 (1926); United States v. Schwarz, 460 F.2d 1365, 1371-72 (7th Cir. 1972); United States v. Ahtanum Irrigation Dist., 236 F.2d 321 (9th Cir. 1956), cert. denied, 352 U.S. 988, 77 S.Ct. 386, 1 L.Ed.2d 367 (1957); Schaghticoke Tribe of Indians v. Kent School Corp., 423 F.Supp. 780, 784-85 (D.Conn.1976).
. See also United States v. Forness, 125 F.2d 928, 932 (2d Cir.), cert. denied, 316 U.S. 694, 62 S.Ct. 1293, 86 L.Ed. 1764 (1942); Schaghticoke Tribe of Indians v. Kent School Corp., supra at 783-84; Narragansett Tribe of Indians v. *631Southern Rhode Island Land Dev. Corp., 418 F.Supp. 798, 804 (D.R.I.1976).
. The defendants question the constitutionality of 25 U.S.C. § 194. In discussing the validity of laws granting special treatment to Indians the Supreme Court emphasized in Morton v. Mancari, 417 U.S. 535, 554-55, 94 S.Ct. 2474, 2485, 41 L.Ed.2d 290 (1974), that:
On numerous occasions this Court specifically has upheld legislation that singles out Indians for particular and special treatment. This unique legal status is of long standing and its sources are diverse. As long as the special treatment can be tied rationally to the fulfillment of Congress’ unique obligation toward the Indians, such legislative judgments will not be disturbed.
(Citations omitted).
See also Moe v. Confederated Salish & Koote-nai Tribes, 425 U.S. 463, 479-80, 96 S.Ct. 1634, 48 L.Ed.2d 96 (1976); Fisher v. District Court, 424 U.S. 382, 390-91, 96 S.Ct. 943, 47 L.Ed.2d 106 (1976). Cf. McClanahan v. Arizona State Tax Comm’n, 411 U.S. 164, 93 S.Ct. 1257, 36 L.Ed.2d 129 (1973); Williams v. Lee, 358 U.S. 217, 79 S.Ct. 269, 3 L.Ed.2d 251 (1959); Simmons v. Eagle Seelatsee, 244 F.Supp. 808 (E.D. Wash.1965), aff’d, 384 U.S. 209, 86 S.Ct. 1459, 16 L.Ed.2d 480 (1966).
. Two cases have cited § 194, United States v. Sands, 94 F.2d 156 (10th Cir. 1938), and Felix v. Patrick, 36 F. 457 (C.C.D.Neb.1888), aff’d, 145 U.S. 317, 12 S.Ct. 862, 36 L.Ed. 719 (1892), but neither case expounds upon the effect the section should be given.
. Section 22 of the 1834 Indian Non-Intercourse Act, Act of June 30, 1834, 4 Stat. 733, upon which 25 U.S.C. § 194 is based, is derived from a similar provision in an 1822 non-intercourse act. See Act of May 6, 1822, § 5, 3 Stat. 683. A 1924 Attorney General’s opinion uses § 194 as an example of a long established practice of safeguarding Indian rights.
From the beginning of its negotiations with the Indians, the Government has adopted the policy of giving them the benefit of the doubt as to the questions of fact or the construction of treaties and statutes relating to their welfare. An illustration of this is found in section 2126 of the Revised Statutes (Act of June 30, 1834, 4 Stat. 733) [25 U.S.C. § 194], which provides:
“In all trials about the right of property in which an Indian may be a party on one side, and a white person on the other, the burden of proof shall rest upon the white person, whenever the Indian shall make out a presumption of title in himself from the fact of previous possession or ownership.”
This practice of safeguarding the Indian has been continuously adhered to. Treaties have been considered, not according to their technical meaning, but in the sense in which they would be naturally understood by the Indians.
34 Op.Att’y Gen. 439, 444 (1925).
. Adoption of findings proposed by the successful litigant will be upheld where supported by substantial evidence and not otherwise clearly erroneous. However, in the present case the entire opinion of the trial court relat-
ing to the evidence and findings of fact is essentially a memorandum written by the defendants. Under the circumstances we feel compelled to repeat the admonition of the United States Supreme Court in United States v. El Paso Natural Gas Co., 376 U.S. 651, 656-57, 84 S.Ct. 1044, 1047, 12 L.Ed.2d 12 (1964):
Those findings, though not the product of the workings of the district judge’s mind, are formally his; they are not to be rejected out-of-hand, and they will stand if supported by evidence. Those drawn with the insight of a disinterested mind are, however, more helpful to the appellate court. Moreover, these detailed findings were “mechanically adopted,” to use the phrase of the late Judge Frank in United States v. Forness, 125 F.2d 928, 942, and do not reveal the discerning line for decision of the basic issue in the case.
(Emphasis added) (citations omitted) (footnote omitted).
. Under Iowa law there is a common law presumption in favor of a finding of accretion. Kitteridge v. Ritter, 172 Iowa 55, 151 N.W. 1097 (1915). No such presumption exists under Nebraska law. See Jones v. Schmidt, 170 Neb. 351, 102 N.W.2d 640 (1960). Whether a presumption of accretion exists under federal law is uncertain; Mr. Justice Douglas alludes to such a presumption in his dissent in Mississippi v. Arkansas, 415 U.S. 289, 295-96, 94 S.Ct. 1046, 39 L.Ed.2d 333 (1974) (Douglas, J., dissenting). Cf. Nebraska v. Iowa, 143 U.S. 359, 369, 12 S.Ct. 396, 36 L.Ed. 186 (1892).
The existence of a presumption of accretion, however, does not affect the outcome here.
*633Under the Federal Rules of Evidence a presumption loses its vitality once sufficient evidence on a disputed issue has been presented to permit a fact finder to act upon it. See Fed.R.Evid. 301; Louisell, Construing Rule 301: Instructing the Jury on Presumptions in Civil Actions and Proceedings, 63 Va.L.Rev. 281, 285 (1977); Sperberg v. Goodyear Tire & Rubber Co., 519 F.2d 708, 713 (6th Cir.), cert. denied, 423 U.S. 987, 96 S.Ct. 395, 46 L.Ed.2d 303 (1975); 1 J. Weinstein & M. Berger, Weinstein’s Evidence fl301[02], at 301-28 (1976); McCormick’s Handbook of the Law of Evidence Ch. 36, § 345 (2d ed. E. Cleary ed. 1972). The Tribe having presented substantial conflicting evidence on the issue of accretion, any presumption of accretion .disappeared and had no further effect on their case.
The presumption of accretion, which affects only the burden of going forward with evidence, should not be confused with the burden of proof, that is the risk of nonpersuasion, found in § 194. As Professor Fleming James noted:
The term “burden of proof’ is used in our law to refer to two separate and quite differ-
ent concepts. . . . The two distinct concepts may be referred to as (1) the risk of non-persuasion, or the burden of persuasion or simply persuasion burden; (2) the duty of producing evidence, the burden of going forward with the evidence, or simply the production burden or the burden of evidence.
James, Burdens of Proof, 47 Va.L.Rev. 51 (1961).
Section 194 is designed to allocate the burden of persuasion and not merely to affect the burden of going forward with the evidence.
. The court observed in its opinion that:
The plaintiffs failed to sustain their burden of proving that any sudden change of the Missouri River channel occurred in the Blackbird Bend area detaching a block of Omaha Indian Reservation land from the Nebraska bank to the Iowa bank, which land was capable of identification as such, either during the period from 1867 to 1879, 1906 to 1923, as contended by the plaintiffs, or at any other time material herein.
433 F.Supp. at 88.
. The actual trial lasted over one month; the trial record constitutes 3,216 pages and includes over 150 exhibits.
. The rule of accretion, which found its origins in Justinian’s Institutes—
Moreover, that ground which a river hath added to your estate by alluvion, becomes your own by the law of nations. And that is said to be alluvion, which is added so gradually, that no one can judge how much is added at each moment of time.
T. Cooper, The Institutes of Justinian, Lib. II, Tit. I, § 20 (3d ed. 1852)— was based on the proposition that the “imperceptible nature of the acquisition” is “too minute and valueless to appear worthy of legal dispute or separate ownership.” Hall, Rights of the Crown in the Sea-Shore, in S. Moore, A History of the Foreshore 793 (1888). See also 1 G. Baker, Halleck’s International Law ch. VI, at § 25 (1908); 8 Op.Att’y Gen. 175, 177-78 (1856). Imperceptibility, in the sense that “though the witnesses may see from time to time that progress has been made, they could
not perceive it while the process was going on,” St. Clair v. Lovingston, 90 U.S. (23 Wall.) 46, 68, 23 L.Ed. 59 (1874), is the accepted test for accretion today. See, e. g., Littlefield v. Nelson, 246 F.2d 956, 958 (10th Cir. 1957); United States v. Commodore Club, Inc., 418 F.Supp. 311, 322 (E.D.Mich.1976); Schafer v. Schnabel, 494 P.2d 802, 807 n. 19 (Alaska 1972).
. Justinian described avulsion as follows:
But, if the impetuosity of a river should sever a part of your estate, and adjoin it to that of your neighbour, it is certain, such part would still continue yours. .
The Institutes, supra § 21.
. The early movements of the Missouri River were described as “remarkably impetuous,” flooding being common on the river.
The regular floods are two in number, and usually occur in April and in June. The first is extremely violent and of short duration, rarely lasting over a week or ten days; it seems to come largely from the upper river.
*635The June rise, although generally higher, is of longer duration, being influenced by local rains and the general saturation of the soil. . The April rise is generally the most destructive, for it shows, for a time at least, a tendency to follow the channels developed during the low water season preceding, and, as a consequence, the banks are attacked with extreme violence. . . . Both, however, have sufficient power to produce tremendous effects and bring about the most astonishing changes.
Preliminary Report Upon the Improvement of the Navigation of the Missouri River, in Annual Report of the Chief of Engineers app. S, at 1651 (1881).
. The perceptible nature of the erosion along the bank was described as follows:
[D]uring each flood there was usually carried away a strip of land from off said river bank from two hundred and fifty to three hundred feet in width, which loss of land could be seen and perceived in its progress; that as much as a city block would be cut off and washed away in a day or two; that blocks or masses of earth from ten to fifteen feet in width frequently caved off and fell into the river and were carried away at one time
St. Louis v. Rutz, 138 U.S. 226, 231, 11 S.Ct. 337, 339, 34 L.Ed. 941 (1891).
. The Court in its concluding paragraph relates:
It appears, however, from the testimony, that in 1877 the river above Omaha, which had pursued a course in the nature of an
ox-bow, suddenly cut through the neck of the bow and made for itself a new channel. This does not come within the law of accretion, but of that of avulsion. By this selection of a new channel the boundary was not changed, and it remained as it was prior to the avulsion, the center line of the old channel; and that, unless the waters of the river returned to their former bed, became a fixed and unvarying boundary, no matter what might be the changes of the river in its new channel.
Nebraska v. Iowa, 143 U.S. at 370, 12 S.Ct. at 400.
. It is in this context that the Nebraska cases relied upon by the district court relate that one of the significant factors of avulsion is identifiable land in place. See Conkey v. Knudsen, 143 Neb. 5, 8 N.W.2d 538 (1943), vacating 141 Neb. 517, 4 N.W.2d 290 (1942); Independent Stock Farm v. Stevens, 128 Neb. 619, 259 N.W. 647 (1935); Iowa R.R. Land Co. v. Coulthard, 96 Neb. 607, 148 N.W. 328 (1914). These and several other state cases — Yutterman v. Grier, 112 Ark. 366, 166 S.W. 749, 751 (1914); Longa-baugh v. Johnson, 321 N.E.2d 865, 867 (Ind. App.1975); Coulthard v. Stevens, 84 Iowa 241, 50 N.W. 983,-984 (1892); McCormick v. Miller, 239 Mo. 463, 144 S.W. 101, 103 (1912); Attorney General ex rel. Becker v. Bay Boom Wild Rice & Fur Farm, 172 Wis. 363, 178 N.W. 569, 573 (1920) — which discuss identifiable land in place as one of the key factors in a finding of avulsion all ultimately rely on either Nebraska v. Iowa or a Missouri case, Benson v. Morrow, 61 Mo. 345 (1875).
. The language and expressions in the Nebraska v. Iowa decision may, as the Supreme Court of Alabama remarked, “cause some confusion unless care is had in observing them.” Greenfield v. Powell, 220 Ala. 690, 127 So. 171, 172-73 (1930). See also Willett v. Miller, 176 Okl. 278, 55 P.2d 90, 93-94 (1935).
. Several other decisions have also recognized that the submergence of land around or over which a channel shifts does not prevent a finding of avulsion. See, e. g., Widdicombe v. Rosemiller, 118 F. 295, 299 (C.C.W.D.Mo.1902); Fowler v. Wood, 73 Kan. 511, 85 P. 763, 768 (1906); Nix v. Dickerson, 81 Miss. 632, 33 So. 490 (1903). Cf. Mulry v. Norton, 100 N.Y. 424, 3 N.E. 581 (1885).
. In Eaton v. Francis, 484 P.2d 128, 131 (Colo. App.1971), the court held that:
According to the trial court’s findings, which were based upon sufficient evidence and are therefore binding upon appeal, the Arkansas River moved from its old location to its new one as result of the great flood in 1921. To gain by accretion, it is necessary to show a slow imperceptible shifting of the river’s course over a long period of time. A sudden violent shift or avulsion, as the court found occurred here, does not shift the boundaries of the lot to conform to the new banks of the river. Rather the lots remain the same, that is, bounded upon the old bank of the river not the new.
(Citations omitted).
See also City of Lawrence v. McGrew, 211 Kan. 842, 508 P.2d 930, 932 (1973); Wood v. McAlpine, 85 Kan. 657, 118 P. 1060 (1911); Fowler v. Wood, 73 Kan. 511, 85 P. 763 (1906); Sharp v. Learned, 195 Miss. 201, 14 So.2d 218, 220 (1943); Bode v. Rollwitz, 60 Mont. 481, 199 P. 688 (1921); Nolte v. Sturgeon, 376 P.2d 616, 619-21 (Okl.1962); Buchheit v. Glasco, 361 P.2d 838, 841 (Okl.1961); Harper v. Holston, 119 Wash. 436, 205 P. 1062, 1064 (1922). Cf. Coastal Indus. Water Auth. v. York, 532 S.W.2d 949, 952 (Tex.1976); Jourdan v. Abbott Constr. Co., 464 P.2d 311, 314 n. 3 (Wyo.1970).
. Both state and federal case law also recognize an exception to the accretion rule where a
stream moves gradually around or jumps across a land area, thereby markedly altering the river’s channel. The rule in these cases, sometimes known as “the island rule,” has not been confined to islands. In Davis v. Anderson-Tully Co., 252 F. 681, 685 (8th Cir. 1918), the court applied the principle to a peninsula and observed:
To the rule stated in this clause there is a well-established and rational exception. It is that when a navigable stream changes its main channel of navigation, not by creeping over the intermediate lands between the old channel and the new one, but by jumping over them or running around them and making or adopting a new course, the boundary remains in the old channel subject to subsequent changes in that channel wrought by accretion and erosion while the water in it remains a running stream, notwithstanding the fact that the change from the old channel to the new one was wrought gradually during several years by the increase from year to year of the proportion of the waters of the river passing over the course which eventually became the new channel, and the decrease from year to year of the proportion of its waters passing through the old channel until finally the new channel became the main channel of navigation.
(Emphasis added).
See also Washington v. Oregon, 211 U.S. 127, 134-36, 29 S.Ct. 47, 53 L.Ed. 118 (1908); Missouri v. Kentucky, 78 U.S. (11 Wall.) 395, 403-11, 20 L.Ed. 116 (1870); Commissioners of Land Office v. United States, 270 F. 110, 113— 14 (8th Cir. 1920), appeal dismissed, 260 U.S. 753, 43 S.Ct. 14, 67 L.Ed. 497 (1922); State v. Ecklund, 147 Neb. 508, 23 N.W.2d 782, 789-90 (1946).
. One author has noted that changes or movements of rivers may be divided into three basic categories: (1) gradual changes in the river caused by erosion and accretion; (2) sudden changes in the river caused by erosion and avulsion without cutting a new bed; and (3) where the river itself cuts a new bed. In discussing these changes, Professor Bouchez explored the policy rules behind boundary river principles:
When an alteration in the boundary river has been caused by a gradual process of erosion and accretion, the best solution is the adjustment of the boundary to the changed situation. Such a procedure will mean that the functional meaning of the thalweg boundary will be maintained. In addition the damage caused to one of the States by slight alterations will generally be of minor importance. Damage arising from slight alterations will not continuously affect in the long run only one of the riparian States.
The second category of alterations is a more difficult problem. Maintenance of the boundary as it existed before the alteration of the Thalweg will abolish the functional meaning of the thalweg boundary. If, on the other hand, the boundary follows the new thalweg serious damage may be caused to one of the riparian States. An example of such a situation is the Chamizal tract. The interests of the State put at a disadvantage by the instantaneous alterations of the thalweg must be considered as the dominant factors in order to find an equitable solution.
If the injured State is primarily interested in navigation the new thalweg is perhaps the
best boundary. If the area, which has been partly destroyed or seriously damaged by the alterations of the thalweg, is of vital importance for a State, the maintenance of the old boundary is to be preferred.
Bouchez, The Fixing of Boundaries in International Boundary Rivers, 12 Int’l & Comp.L.Q. 789, 808 (1963).
. The statement in Bonelli followed from earlier authorities which similarly emphasized the inequities that would arise were the rule of avulsion not available.
[Ijf the change does not come within the definition of gradual and imperceptible, but the land is rapidly washed away on one side and formed on the other, the fiction should give way to the fact, and the owner should not lose title to his property. The title to the land itself is of more importance than the riparian right of access to the water or convenience of having a natural, rather than a mathematical, boundary; and rules which were made for convenience should not be permitted to wrest the title to land from its true owner.
3 Farnham on Waters § 848, quoted in Fowler v. Wood, 73 Kan. 511, 525, 85 P. 763, 768 (1906).
. Another form of avulsion, of course, is recognized in the highly unusual case where identifiable land is visibly torn from one bank and carried downstream to a resting place. The Supreme Court, however, in Nebraska v. Iowa, noted that this type of avulsion could not occur on the Missouri River.
*639No engineering skill is sufficient to say where the earth in the bank washed away and disintegrating into the river finds its rest and abiding place. The falling bank has passed into the floating mass of earth and water, and the particles of earth may rest one or fifty miles below, and upon either shore.
143 U.S. at 369, 12 S.Ct. at 399. Cf. St. Louis v. Rutz, 138 U.S. at 249-51, 11 S.Ct. 337.
. The trial court did, however, express a caveat near the end of its opinion, that “certain findings of fact, even if agreed upon, would not lead in the minds of all parties concerned to a definite legal conclusion; i. e. that either an avulsion or an accretion took place at key times in history at particular places on the Missouri River.” 433 F.Supp. at 89. The district court nonetheless concluded that it was “convinced that one particular set of conclusions can be squared with the evidence far better than other proposed conclusions.” Id. Notwithstanding its rejection of the Tribe’s or the government’s broader approach to the doctrine of avulsion, the court conceded that it was “apparent that the movements of the Missouri River have not been so clean and precise that they easily fall into the legal categories conveyed by the terms ‘accretion’ and ‘avulsion.’ ” Id.
. Captain J. F. Friedkin of the United States Corps of Engineers from 1942 through 1944 conducted one of the pioneering studies of river hydrology.
. Defendants’ expert witness, George R. Hall-berg, Chief of the Research Division of the Iowa Geological Survey, testified that every meander bend has a maximum width. Once the river reaches the maximum width it becomes more efficient for it to flow over the inside of the bend, and erode across the point bar, thereby decreasing the width of the bend.
This process occurs at constant discharge rates, and accelerates at high discharge. Hall-berg theorized that some time between 1867 and 1875 Blackbird Bend reached its maximum width and the thalweg began flowing against the meander lobe.
. In 1875 there was a discharge of 140,000 cubic feet per second. In the next four years, the maximum discharges recorded were 300,-000, 200,000, 215,000 and 220,000 cubic feet per second.
. Barrett described the area in 1867 as follows:
Fractional Township 24 N.R. 11 East, the 6th Principal Meridian, is a low, sandy point, subject to frequent inundations from the Missouri River.
Except a few small cottonwood trees and some willows, there is almost no vegetation upon it.
The Missouri River is constantly changing its banks, so that no permanent corners can be established near the water; indeed, except where there are bearing trees, none of the corners in this Township will probably remain longer than the first high freshet in the Missouri.
Small quantities of coal were deposited in the several mounds as per instructions, but the sandy soil will not prevent it from being washed away.
. The 1890 reference is confusing and in obvious error. The area where bar C was located in 1879 is depicted on the 1890 map as “cleared” land.
. The government urges that the vegetation appearing on bar C is inconsistent with the theory of accretion since, if the land had in fact been washed away, it would have been the last of the. visible bars to have formed and the last to develop vegetation.
. Each of defendants’ experts viewed the existence of identifiable land in place as an essential factor in defining avulsion. Raymond L. Huber, a supervisory civil engineer employed by the Corps of Engineers, stated that
avulsion is the transfer of the piece of land from one bank of the river to the other bank of the river, which occurs in such a manner that you can follow the movement of that transfer of land and identify it as the identical land which formerly existed on the other bank, beyond any power of question.
John F. Kennedy, a professor in the Division of Energy Engineering at the University of Iowa, and director of the Iowa Institute of Hydrologic Research, a division of the University of Iowa College of Engineering, defined avulsion as “the river shifting from one side of a block of land to the other leaving the intervening land undisturbed in the process.” Finally, Dr. George R. Hallberg testified that:
Avulsion is a term I don’t use technically. But from much of my work, what I understand avulsion to mean is a very sudden perceptible movement which would cut off and abandon a river and leave the land contained by that old channel essentially intact.
. Wigmore in his treatise on evidence notes that while.one inference may properly be built on another inference, 1 J. Wigmore, Evidence § 41 (3d ed. 1940), there is a limit to proof offered in this manner. As an example of the correct method of treating inferences Wigmore quotes from the case of New York Life Ins. Co. v. McNeely, 52 Ariz. 181, 79 P.2d 948 (1938), in which the court observed:
“It is true, of course, that in everyday life, all men frequently act as the result of the repeated piling of inferences upon inferences, and, as a matter of strict logic, if an inference has any probative value whatever in aiding one to determine the ultimate question of fact, it should be considered. The principle which is applied by the average man in his own private affairs usually is that no matter how many inferences are piled on each other,
it is only necessary that each successive inference should be more probable than any other which might be drawn under all the circumstances. The Courts, however, have always insisted that the life, liberty and property of a citizen should not be taken away on possibilities, conjectures, or even, generally speaking, a bare probability.”
1 Wigmore, supra § 41, at 439.
. Cf. Logsdon v. Baker, 170 U.S.App.D.C. 360, 361, 517 F.2d 174, 175 (1975), where the court held that reliance on “brake marks” purported to be visible in photographs but which could not be discerned by the district judge or the appeals court was held to be an inadequate basis for an expert opinion on the cause of an accident since the expert’s opinion would be “too speculative.”
. The Tribe and the government substantiated their argument that the remnant channels were more consistent with avulsive than accretive movement by presenting evidence of soil samples taken in the area immediately to the east of the Barrett Survey line which showed concentrations of silt and clay in the remnant channels. Concentrations of silt and clay were said by plaintiffs’ experts to be inconsistent with the gradual, imperceptible, and progressive action associated with accretion which was said to produce more uniform deposits. The defendants attempted to refute this by testimony that the remnant channel was cut off at the top by formation of a point bar and alluvial deposits would therefore not reach the channel.
. See Galloway v. United States, 319 U.S. 372, 386-87, 63 S.Ct. 1077, 87 L.Ed. 1458 (1943). Raymond L. Huber, one of defendants’ witnesses, conceded that, given the factual basis available to him, he was only able to give an “educated guess” as to what caused the river to move between 1875 and 1879.
. The limiting width of a meander bend is the maximum distance the river will travel away from the general direction of a river’s flow. *644See J. Friedkin, Meandering of Alluvial Rivers 14 (1945).
. A chute channel is defined as “a channel across a bar or a pointway channel. It differs from a cut-off, wherein a river cuts through a narrow neck which has been developed between the upper and lower arms of a bend.” Id. n. 1.
. Dr. Kennedy, defendants’ expert, verified McQuivey’s testimony in observing that “[during these extreme inundations, the channel, of course, likes to take a shorter path.”
. The fact that during high water a river’s flow tends to move to the inside of a meander bend was substantiated by the defendants’ witness Huber who testified:
I will show you an example of a chute cutoff. Take the example of Tieville Bend, which is just below the Blackbird Bend area, below the Barrett Survey—
Now, there was a shore chute which had never entirely dried up, had some flow, so that in the high water period of 1943 and later again in 1952 the velocity across this area from the upper end to the lower end of the chute was so great as compared with the velocity around the old bend that it caused
some filling of the old channel and it caused a new channel to develop.
Essentially this channel filled up at the upper end to the point that I believe you can walk across it. The lower end is still open, so this would be a chute avulsion.
Huber also pointed out that:
In a very low stage there is insufficient hydraulic energy for the river to cut across any sandbars with the velocity of two or three miles to an hour — two or three miles an hour, I said, so that it has to follow the easternmost high bank. Then as the stage increases up to flood stage, which could be even up to ten miles an hour, a channel will then seek the shorter path across the area, across the bare peninsula, and develop a new channel west, which will be even deeper than the channel against the main bank, for the reason that you have greater bed scar [sic; should read scour], you have greater discharge.
. McQuivey pointed out several remnant channels in the area of the peninsula.
. In United States v. Flower, 108 F.2d 298 (8th Cir. 1939), the court found a neck cut-off had occurred only a few miles above Blackbird Bend in 1916 after the river had formed a very wide meander bend.
. Friedkin provides the following illustration of the circumstances which promote natural cut-offs.
DEVELOPMENT OF NATURAL CUT-OFF
[[Image here]]
Friedkin, supra plate 25.
An examination of Tribe Exhibit 96, an 1875 map of Monona County, Iowa, shows a narrowing of the meander bend in the area between sections 29 and 30 which closely resembles Friedkin’s sketch.
. In the years immediately prior to 1879 the north and south banks of the Blackbird Bend meander lobe had considerably narrowed. Noting this condition, Dr. Robinson testified that:
[F]rom a study of river morphology we can predict that something is about to happen. The river is out of balance with its environment and it’s because of this resisted layer at the south that does not allow the classic of the wood [sic] where we have homogeneous
materials throughout the river to move south.
I would predict if I were in a study that there would be a point bar cut-off or a shoot [sic] developed through there that would develop into the main channel of the river and we would have an abandonment of the 1875 channel and the development of a new channel on a shorter and straighter course between the two limbs of the meander.
. Studies made by the Corps of Engineers received in evidence also lend support to the concept that channels often change location precipitously in both high and low stages. The report reads:
During flood stages the general movement of sand is in a measure arrested on the shoals, and, as a consequence, the low-water channels leading through them are silted up, while new ones are developed which accommodate the high-water flow. As the river falls these latter channels are silted up, and the water ponds up behind the bar until sufficient head is attained to force a breach through the crest at one or several points.
During the course of the low-water season the main channel keeps shifting from one subsidiary channel to another; these changes of location are often very great, and the uncertainty thereon attendant forms one of the greatest drawbacks to navigation. The high-water channels, before alluded to, are, in the Missouri, of the same general character as those of low-water, but they rarely coincide with them. They are less tortuous as a rule, but not of much greater depth, and they are liable to the same rapid changes of location.
*646Preliminary Report Upon the Improvement of the Navigation of the Missouri River, in Annual Report of the Chief of Engineers app. S, at 1655 (1881).
. Assuming that the 1890 map shows stabilized accretion land, as the district court found, it is plausible that the “accretion” may have formed from deposition of sediment resulting from a sudden and perceptible shift of the thal-weg.
. The testimony related to the existence of trees as proof of identifiable land in place was conflicting. Two trees pre-dating the 1912 movement were identified by the Tribe’s witness George S. Gorsuch, a forester with extensive experience in dendrochronology (the study of time-span within trees). The defendants contended, and the trial court found that these trees were located on the edge of a sandbar that had allegedly built south of the northerly high bank prior to 1912. However, the contours of the sandbar were never surveyed and its exact location could not have been ascertained.
. The allotment letters concerned requests by certain members of the Omaha Tribe that their land allotments be exchanged for new allotments because the land originally allotted them was being or had been washed away by the Missouri River.
. The parallelism of the bars between the remnant channel formations was also relied on to support an accretion theory. We have discussed the speculative significance of the parallelism of land features earlier. See discussion in section VI(C) supra.
. Hallberg stated that the existence of marshy areas and well-defined channel remnants to the east, in contrast to drier areas and less well-defined remnant channels to the west, supported the theory that as the channel shifted away from its bank to keep up with the growing point bar it subsequently backfilled the area creating the remnants.
. The testimony of Ross Willey indicated that the river still flowed in the easterly English Bayou area in 1916. Another lay witness, Judge George W. Prichard, who had traveled the area on horseback in 1919, gave testimony which tends to support the fact that most of the southerly movement of the river had been completed by 1919.
. A large stand of timber shown in the northwest corner of the Barrett Survey area prior to 1923 was no longer visible in a 1927 aerial survey of the area, indicating the land on which it stood had been eroded.
. The absence of trees in the area north of the Barrett Survey does not support an inference that the land had been completely eroded by the river’s southwesterly movement. Depending upon the nature of the river’s northerly movement, all trees growing in the area prior to 1912 may have been washed away. In addition, the record indicates that many of the trees growing in the area after 1923 were cut for firewood in the 1930’s.
. The government had performed extensive soil analysis of the Blackbird Bend area. On government exhibit 151, which summarized the results of that analysis, the government’s experts depicted those areas within the Blackbird Bend area where the soil samples indicated remnant channels, and those areas where the soil samples indicated bar-like formations.
. These soil samples point out the existence of the remnant channels. McQuivey, the government’s expert, testified that:
Looking at Sample 61 and 56, 61 was silt on the top. There was some fine, very fine sand down to a depth of four feet, and then down to a depth of approximately twenty feet was soil clay. That is an indication of an old channel-type deposit, where the water has not been running in, and it’s an area of back water or slack water to where the silts and the fine clays can be settled out.
So basically, we can determine from the soil samples if there was an old channel there.
The defendants’ witness did not dispute the validity of the soil samples but, instead, presented only alternate theories as to how possibly the non-alluvion soil was deposited.
. How the river reached its outer eastern limit in 1875 or moved to its most northerly point in and around 1912 is outside the confines of this case. The Tribe oontends erosion and accretion were responsible for these movements and there exists statements in the briefs in which it appears that the defendants have conceded this to be true. However, any such concession may have been predicated on an erroneous view of the law and should not be deemed binding under the circumstances. In using the word “accretion” in the context of the easterly movement from 1875 to 1879 and of the northeasterly movement from 1890 to 1912 we understand the defendants as merely saying that there has been deposition of land added to the Barrett Survey area. This land is all outside of the Barrett Survey and beyond the limits of the reservation established by treaty in 1854. The application of § 194 placing the burden of proof on the defendants in the present litigation is not controlling as it affects the area outside the original reservation. We do not in any way attempt to prejudge the several claims; nor do we foreclose the right of the plaintiffs to urge that § 194 is applicable under different proof. We simply note that the same proof showing presumptive title (Treaty of 1854) to the reservation cannot govern any future litigation concerning the lands outside that area.
. The government excepted from its complaint any claim to approximately 400 acres of land within the Barrett Survey which may have been allotted to individual Indians and subsequently patented to non-Indians. Any current claims to those lands by the Tribe might be affected by issues of adverse possession and laches. We therefore remand to the district court those claims related to the land excepted from the government’s complaint for a determination on the above mentioned issues.
. Plaintiffs question whether the trial court’s judgment was final since it lacked certification under Fed.R.Civ.P. 54(b). The claims here on appeal were severed from claims to land outside the Barrett Survey, asserted in C-75 — 4067, by order of Judge McManus. The severed claims in C-75-4067 had been previously joined with quiet title actions pertaining only to the 2,900 acres of land within the Barrett Survey filed by the United States in C-75-4024 and by the Tribe in C-75-4026 which are on appeal here. Thus, the district court settled all of the questions of ownership involving the lands within the Omaha Indian Tribe’s original 1854 reservation boundaries. Under the circumstances we view the severed claim in C-75-4067 as a wholly separate action which had been joined with the quiet title suits so that the judgment entered by the district court resolved the entire controversy then before it. Rule 54(b) therefore, has no application since all, not “fewer than all of the claims,” before the trial court were resolved by its judgment.