*97OPINION OF THE COURT
Plaintiffs appeal from a judgment of the district court refusing to reapportion the Virgin Islands’ fifteen-member legislature. In this appeal, as in the district court, they assert that the district of St. Croix should be entitled to elect a majority of the territorial representatives.
I
On May 18, 1972, the Virgin Islands Senate, the territory’s unicameral legislature, enacted an apportionment plan, Act No. 3221, 2 V.I.C. §§ 101-03. That act, presently in force, divides the territory’s three islands into two legislative districts, one consisting of St. Croix and the other consisting of St. Thomas and St. John. Fourteen senators are divided equally between the two districts while the fifteenth, known as the senator at large, is elected by the territory’s entire population. The senator at large, however, must be a resident of St. John.1
In enacting this plan, the Senate relied upon gross population figures drawn from the 1970 Census. These figures credited St. Croix with a population of 31,779, St. Thomas with a population of 28,960, and St. John with a population of 1,729. The district of St. Thomas-St. John thus included 30,689 persons, or 1090 persons less than the district of St. Croix.
Plaintiffs, a resident of St. Croix and six of St. Croix’s seven senators, brought this class action seeking legislative reapportionment. They assert that St. Croix’s status as *98the more populous of the two legislative districts entitles it to majority representation in the Virgin Islands Senate. Plaintiffs base their claim on the fifth amendment to the Constitution of the United States and on 48 U.S.C. § 1561, which guarantees equal protection of the laws to citizens of the Virgin Islands. After receiving the parties’ evidence, the district court found that St. Thomas-St. John, although perhaps less populous than St. Croix, included more citizens and more registered voters than the latter district. Furthermore, the district court rejected plaintiffs’ contention that the present apportionment gave a majority of the representatives to St. Thomas-St. John. Instead, the district court found that the senator at large represented the entire territory and not just St. Thomas-St. John. On the basis of these findings the district court entered judgment in favor of defendants.
II
On appeal, plaintiffs allege that the district court erred (1) in approving the plan on the basis of data other than that considered by the Senate in 1972; (2) in concluding that the senator at large represents the entire territory rather than just the district of his or her residence; (3) in refusing to give the residents of St. Croix a full eight votes in the Senate rather than seven votes and one senator at large; and (4) in admitting various evidence at the trial. Plaintiffs do not seek to alter either the districting of the Virgin Islands or the number of seats in the Senate. We must consider this appeal, therefore, within the framework of a two-district, fifteen-member plan. Cf. Sixty-Seventh Minnesota State Senate v. Beens, 406 U.S. 187, 197 (1972) (whenever possible courts must defer to a legislature’s choice as to its own size).
*99A.
The legislative journal for May 18, 1972, clearly shows that in apportioning the territory the Senate considered only gross population figures from the 1970 Census: St. Croix, 31,779, St. Thomas, 28,960, and St. John, 1,729. Plaintiffs place exclusive reliance on these figures to demonstrate malapportionment. For the purposes of this appeal, we will assume that the 1972 plan must stand or fall on the basis of these figures.
B.
Plaintiffs contend that the district court erred in concluding that the Senate is evenly apportioned 7-7-1 rather than 8-7 in favor of St. Thomas-St. John. They point out that the senator at large must reside in St. John, which is included in the smaller of the two legislative districts.
In Fortson v. Dorsey, 379 U.S. 433, 438 (1965), an early reapportionment case, the Supreme Court held that “[i]t is not accurate to treat a senator from a multi-district county as the representative of only that district. . . wherein he resides.” In that case the plan of apportionment required various state senators to reside in particular subdivisions of the voting unit. Nevertheless, each senator’s tenure depended “upon the county-wide electorate [and] he must be vigilant to serve the interests of all the people in this county....” Id.
The Supreme Court has guarded Fortson jealously. Several lower courts have attempted to distinguish the facts in that case from various apportionment plans that included sub-district residency requirements. Each time, the Supreme Court has rebuffed the challenge. See Clark v. Peters, 422 U.S. 1031, vacating mem., 508 F.2d 267 (5th Cir. 1975); Dallas County v. Reese, 421 U.S. 477 (1975) (per curiam); Dusch v. Davis, 387 U.S. 112 (1967). In Dallas County v. Reese, supra, at 481, the Court completely *100discredited the “theoretical presumption . . . that elected officials will represent the districts in which they reside rather than the electorate which chooses them.”
Admittedly, the Supreme Court has never faced a plan of apportionment like that under scrutiny here. The Virgin Islands have only two legislative districts; the single senator at large may act decisively when the other senators split seven to seven along district lines. Nevertheless, none of the cited cases indicates that such a configuration would defeat Fortson’s general rule. We believe that the senator at large must be counted as a representative of the entire territory and not as a representative of St. Thomas-St. John alone.
We are not called upon to decide whether a different result would be dictated had plaintiffs proven that the senator at large was, in fact, a “front” for the interests of only one district. See Dallas County v. Reese, supra, at 480; Dusch v. Davis, supra, at 116. The district court found the record “totally devoid” of evidence that might suggest such a charade. Plaintiffs do not challenge this finding on appeal.
We conclude that the present plan of apportionment must be characterized exactly as the statute itself would suggest: seven senators representing St. Croix, seven senators representing St. Thomas-St. John, and one senator representing the entire territory.
C.
Our conclusion that representation in the Virgin Islands is divided evenly between the two districts does not dispose of this appeal. Plaintiffs argue that, even given such a characterization, the district of St. Croix remains unconstitutionally underrepresented. The majority of the population, they assert, is entitled to control the majority of the legislature.
*101Initially, we note that traditional measures of under-representation do not mandate reapportionment in this case. Assuming, as we must, that the representational scheme is 7-7-1, the district of St. Croix, when compared with St. Thomas-St. John, is underrepresented in the Senate by no more than 3.5 per cent.2 The Supreme Court has recognized an informal cut-off point of 10 per cent for challenges to apportionment of legislative bodies other than the Congress of the United States. See Connor v. Finch, 431 U.S. 407, 418 (1977). See also Gaffney v. Cummings, 412 U.S. 735 (1973); White v. Regester, 412 U.S. 755 (1973). The deviation in this case falls comfortably below that threshold.
Plaintiffs’ argument, however, transcends mathematics. They flatly assert that any majority, no matter how small, entitles them to control a majority of the Senate. They purport to draw this rule directly from the seminal case of Reynolds v. Sims, 377 U.S. 533 (1964). There, the Supreme Court held that state legislatures must be apportioned on the basis of population. In considering Alabama’s plan of apportionment, which allowed 25 per cent of that state’s population to elect majorities in both houses of the legislature, the Supreme Court decried governance by a minority of the electors:
Logically, in. a society ostensibly grounded on representative government, it would seem reasonable that a majority of the people of a State could elect a majority of that State’s legislators. To conclude differently, and to sanction minority control of state legislative bodies, would appear to deny majority rights in a way that far surpasses any possible denial of minority rights that might otherwise be thought to result.
*102Id. at 565. Later in the same opinion, the Court returned to this theme.
Our constitutional system amply provides for the protection of minorities by means other than giving them majority control of state legislatures. And the democratic ideals of equality and majority rule, which have served this Nation so well in the past, are hardly of any less significance for the present and the future.
Id. at 566.
Plaintiffs cite these expansive passages and suggest that this court must vindicate the “democratic ideal” of majority rule by providing St. Croix with a majority of the fifteen votes in the Virgin Islands Senate. In another passage, however, Reynolds sets forth the “overriding objective” of legislative apportionment: “substantial equality of population among the various districts, so that the vote of any citizen is approximately equal in weight to that of any other citizen . . . .” Reynolds v. Sims, supra, at 579. In many cases, as in Reynolds, pursuit of this polestar will allow districts containing a majority of the population to elect a majority of the representatives. But where two evenly balanced districts choose a small number of representatives, we perceive a tension between “substantial equality” of voters and “majority rule” as defined by plaintiffs.
The data in this case illustrate the tension. In 1970, St. Croix claimed 50.9 per cent of the territorial population; St. Thomas-St. John claimed the remaining 49.1 per cent. As already noted, this small disparity results, under the 7-7-1 plan of apportionment, in 3.5 per cent underrepresentation of St. Croix. If plaintiffs enjoyed an eight-to-seven majority in the Senate, however, St. Thomas-St. John would be underrepresented by approximately 9.9 per cent.3 By granting plaintiffs’ requested relief we would *103create a greater disparity in representation than already exists.
Forced to choose between the “overriding objective” of equality of franchise and plaintiffs’ notion of majority rule, we believe that the present plan of apportionment is more consistent with the requirements of Reynolds v. Sims. Given our reliance on plaintiffs’ gross population figures, we need not decide whether the distribution of citizens or of registered voters offers a permissible basis for apportionment. See Burns v. Richardson, 384 U.S. 73, 92-95 (1966).
D.
Plaintiffs contend that the district court erred in admitting the testimony of Kenneth Callwood, defendants’ demographic expert, and in soliciting an affidavit from the Virgin Islands’ Supervisor of Elections, a defendant in this action. Because neither we nor the district court relied upon Callwood’s testimony in reaching our conclusions, plaintiffs were not prejudiced by its admission. Similarly, we have not relied upon the Supervisor’s affidavit, which dealt with voter registration in the two districts. The district court’s solicitation and use of that affidavit, although inconsistent with the Code of Judicial Conduct’s prohibition against ex parte contacts, does not constitute reversible error.
III
The order of the district court entering judgment in favor of defendants will be affirmed.