dissenting:
While I entirely agree with the legal standard applied by the majority in this case, I disagree with its assessment of the facts before this court. I believe that an analysis of all of the facts presented to the district court makes clear that the residents of the City of Brewton do not have an interest in the operation of the Escambia County school system substantial enough to permit them to vote in the county school board elections. Accordingly, I must dissent.
I. LEGAL AND FACTUAL CONCLUSIONS IN THE DISTRICT COURT.
As the majority correctly holds, this case is controlled by three cases binding on this court. Creel v. Freeman, 531 F.2d 286 (5th Cir.1976), cert. denied, 429 U.S. 1066, 97 S.Ct. 797, 50 L.Ed.2d 784 (1977); Phillips v. Andress, 634 F.2d 947 (5th Cir. Unit B Jan. 1981); Hogencamp v. Lee County Board of Education, 722 F.2d 720 (11th Cir.1984). The district court in this case reviewed these three opinions and decided that the Creel case articulated a different standard than that applied in the Phillips and Hogencamp eases; the district court rejected as incorrect the standard set out in Phillips and Hogencamp. Dist.Ct.Op. at 11-12. The majority here correctly rejects this differentiation, and views the standard applied in all three cases as consistent:
The proper standards to apply in this case have been delineated in a series of cases before this Court. The party seeking to exclude city residents from voting in the county school board elections has the burden of demonstrating that the application of the Alabama statute here is irrational or wholly irrelevant to the state’s objective of electoral participation in the selection of county school board members. The test for whether the statute is irrational as applied to the particular county is whether the city residents have a substantial interest in the operation of the county school system. If the city residents do not have a substantial interest, then the state must exclude the city residents from voting.
Majority Op. at 772 (citations omitted).1
*776The crucial error made by the district court (and by the majority) in this case, however, is not the incorrect legal standard, but the exceedingly limited findings of fact. It is true that none of the findings of fact made by the district court are erroneous. Indeed, the vast majority of them are drawn directly from the pre-trial order, prepared and agreed to by the parties. The mistake of the district court is that its findings omit or overlook a vast amount of data very relevant to the application of the Creel standard. The majority opinion, for instance, is correct in saying that the “district court’s findings did not contain any conclusions about the relative financing of the systems.” Majority Op. at 774. The district court’s findings, however, ignore extensive testimonial and documentary evidence detailing the source and distribution of the finances of the Escambia County school system. In the face of such limited findings of fact, it is necessary to undertake a thorough and detailed consideration of the evidence presented to the district court.2 This consideration, set out below, leads to the conclusion that the residents of Brewton do not have an interest in the operation of the county schools substantial enough to permit the vote dilution that, as the data demonstrates, is present in Escambia County.
II. THE FACTS OF THIS CASE.
Among the factors considered in the Creel, Phillips, and Hogencamp cases are the amount of student crossover into the county schools, the presence of special educational programs in the county school system, and the financial dependence of the county schools on the cities. These and other factors should be carefully reviewed in deciding this case.
A. Student Crossover.
Both the county and city school boards maintain open door policies, allowing any school-age Alabama resident to attend their schools. Practically speaking, of course, this student crossover usually occurs only between adjacent school districts such as Brewton and Escambia County. The details of this crossover reveal, however, that relatively few Brewton students attend county schools.
1. Regular School Attendance.
Forty-four city residents attend the county schools. Majority Op. at 773; Dist.Ct.Op. at 4. What is left unsaid by the district court is that the county schools have over 5,700 students, and thus the 44 city residents comprise less than 1% of the students in the county school system. Because (in part) of these 44 city students in the county schools, the defendants argue that the City of Brewton residents have a substantial interest in the operation of the county schools. Forty-four students out of 5,700, however, seems much closer to the *777situation in Phillips v. Andress, where the “few” crossovers did not create a significant interest in the county schools. 634 F.2d at 950.
Both the majority and district court opinions note that the crossover in the other direction, from county to city, is 330 students. Thus, of the 1,500 students in the city schools, over 20% of them reside outside of the city. Contrary to the arguments of the defendants in this case, this would suggest that if any group has an interest in the other’s schools, it would be the county residents who have an interest in the city schools. County residents, however, are not permitted to vote to elect the Brewton City Council, which appoints the city board of education.
Viewed from an angle somewhat more favorable to the defendants in this case, the 44 city residents attending county schools represent about 3.4% of all school-age city residents. Thus, a small but not completely insignificant percentage of city children are being educated in the county. Nevertheless, this percentage, and the small absolute number it represents, does not, for me, rise to the level of creating a substantial interest.
Perhaps even more important than the small crossover involved is the applicability of the concept, drawn from another area of the law, that “school officials are under no duty to adjust for the purely private acts of those who chose to vote with their feet.” Davis v. East Baton Rouge Parish School Board, 721 F.2d 1425, 1435 (5th Cir.1983). It is my view that the parents of the 44 city residents who attend county schools have “voted with their feet.” For whatever reason, those parents have chosen to send their children to the county schools rather than the city schools over which they rightly have voting influence. The voluntary choice of this small group does not provide a sufficient justification to allow the 6,600 Brewton residents voting influence in the county school system.3
2. Handicapped and
Retarded Students.
The district court lists, as one component of the interest city residents have in the county schools, the cooperation between the two school systems in providing programs for the multi-handicapped and the trainable mentally retarded students. The county maintains a school for one group, while the city maintains facilities for the other. Yet, as important as these programs may be, they involve only 5 or 6 students in each school. Record, Vol. 2, at 13. As with the 44 students in the regular schools, I do not consider these numbers sufficient to create a substantial interest.
3. Summer School.
Another factor mentioned to support a substantial interest is the fact that the county maintains a summer school system, while the city does not. The attendance figures presented by the court below, however, are somewhat misleading. Fifty-one of the 88 students (in 1984) who attended one summer school run by the county were from the city. Dist.Ct.Op. at 4. This ignores, however, the fact that the county runs more than one summer school, and that city residents attend only the school closest to the city. Record, Vol. 2, at 26-28. The record does not reflect the total number of summer students in schools run by the county. Thus, it is impossible to know what percentage of the summer school students the 51 city residents comprise. Of course, the county board of education, elected in part by the city residents, controls and operates the county’s entire summer program, hot just the one school attended by city residents.
*7784. Vocational Center.
Pursuant to a joint venture agreement, the county and city share a vocational center. Numerically speaking, this program represents the city’s largest group of students who crossover to attend county classes. Over the five years preceding the district court’s ruling, an annual average of 83 city residents attended classes at the center, representing about 43% of the total enrollment at the center.
5. Overall.
Forty-four students, or less than 1% of the county’s total enrollment, simply does not suggest a substantial interest of the city in the county schools. Even including all of the special programs, the city crossover is less than 4% of the county’s enrollment. Yet, as discussed in the following section, the special programs are conducted pursuant to voluntary agreement between the city and the county; these agreements are of the type explicitly discussed in the Creel opinion as “rendering electoral participation in the selection of county school board members by residents of the city school districts unnecessary.” 531 F.2d at 289 (emphasis in original). Whether or not the special programs are considered in assessing student crossover, the small numbers seen here do not begin to create the substantial interest needed to uphold the challenged voting scheme.
If the crossover of 44 students is held to be sufficient to create a substantial interest by the city in the county schools, the effect of the court’s ruling today may be to encourage Escambia County, and counties across Alabama, to close their open doors. If an open door policy and some neighborly cooperation on special programs gives rise to a loss of voting control by the non-city residents of a county, this court is creating a strong impetus against such cooperation.
B. Shared. Responsibility and Control.
The city and county school boards are “two separate boards having responsibilities to two separate school systems. The boards are made up of different people, occupy different offices, employ their own administrative staffs, operate under their own budgets, and have exclusive authority over their own personnel.” Dist.Ct.Op. at 3. Notwithstanding this independence, the defendants argue that the city school board and city residents depend on various services provided by the county board, and thus the city residents should be able to vote in the county board elections. An analysis of the county programs utilized by city residents reveals, however, that the city has a quite adequate voice in the management of the programs, and does not need electoral influence.
1. Vocational Center.
The vocational center, operated as a joint venture between the two school boards, was created pursuant to an “Agreement for Joint Venture.” Plaintiffs’ Exhibit D. Unlike in the Creel case, where the shared vocational center was “under the exclusive control” of the county school board, 531 F.2d at 287, the joint venture contract gives exactly equal control and responsibility for the vocational center to the county and the city school boards.
It is true that the city school board has informally withdrawn from full participation in the operation of the school. Nevertheless, the city has the fully equal contractual right to influence the vocational center, and the city’s voluntary decision to forego that right cannot be used to explain why the city residents need to vote in the county elections. If the city wants to influence the vocational center, it can do so under the contract, without needing the vote.4
In another important respect, the vocational center is quite unlike that in the *779Creel case. In Creel, the city had contributed $212,500 toward the construction of the vocational center, and thus the Creel Court reasoned that the city had made a “substantial investment” in the center. 531 F.2d at 287, 289. In this case, the situation is exactly the opposite. The establishment of the school was financed by state funds, but the county contributed 15 acres of land and a school building. Plaintiffs’ Exhibit D at 2. This contribution by the county was worth over $300,000. Record, Vol. 2, at 31. As far as the record reveals, the City of Brewton made no financial contribution toward the establishment of the vocational eenter. Thus, unlike in Creel, the city has made no “substantial investment” in the center.5 Whatever interest the city has in the center is fully satisfied by the absolutely equal control it can exercise over the center’s operation.
2. Handicapped and Retarded Programs.
The two school boards, by informal agreement, have divided responsibility for programs for multi-handicapped and trainable mentally retarded students. The city board operates a class for the multi-handicapped, while the county serves the trainable mentally retarded. The district court also noted that the county provides transportation for the city students in these classes.
While not governed by a formal contract such as the joint venture agreement controlling the vocational school, the city clearly has sufficient bargaining power to influence the operation of the county run school for the retarded. Presumably, neither the county nor the city want to operate both programs, and thus they are equally dependent on each other for the programs. As for the transportation, the county transports not more than ten city residents. Plaintiffs’ Exhibit A at 2-3.- Neither the existence of these programs, where both jurisdictions are equally dependent on the other, nor the transportation of ten students, is sufficient to create the requisite substantial interest.6
3. Summer Schools.
Unlike the vocational and handicapped programs, the city residents do not, as far as the record reveals, have equal control over the summer school classes, which are operated solely by the county. The most direct power the city residents wield may be in the $30 fee that each student pays the county for the summer classes — it is possible that if the city residents withdrew from the county program, the county might be forced to close the one summer school that services Brewton residents.
Even conceding the limited control Brew-ton residents have over the summer program, the voting influence over the entire county school system still does not seem justified. If the city residents are unhappy with or feel too dependent on the county summer program, I would think that the proper response would be to lobby the city school board for a satisfactory program. And, because the city residents can in fact influence the operation of the vocational programs, the desire for influence over the summer program, combined with the small amount of student crossover, does not alone create a substantial interest.
Beyond crossover and control, the Creel standard considers finances. While the district court did not make detailed findings about the finances, the record reveals only a limited contribution by the city to the county system.
C. Financial Support.
The district court failed to make detailed findings about the financial inter-relationship between the city and the county, although the three precedent cases extensively discussed financial considerations. See Hogencamp, 722 F.2d at 722; Phillips, *780634 F.2d at 951; Creel, 531 F.2d at 288. The record of the case, however, does reveal a great deal of information about the finances of the county schools. The data, especially when compared with that found in Hogencamp and Phillips, indicates that the city residents have only a minimal financial interest in the county school system.
In Hogencamp, the city residents contributed 2.74% of the county school systems budget, an amount explicitly ruled to be “insufficient to create a substantial interest in the city residents.” 722 F.2d at 722. While the Phillips opinion did not designate a specific percentage, it is clear from the opinion that the city residents in that case contributed 1% to 5% of the county school budget; this amount was found to be insufficient. 634 F.2d at 951.
There are two ways that the City of Brewton could contribute to the county schools: direct subsidies or indirectly through tax dollars that flow to the county from the city. In this case, without detailing any of the subsidies, the defendants admitted that any subsidies that might exist amount to no more than 3% of the county budget. Plaintiffs’ Exhibit A at 2. This “3% or less” figure is very close to the 2.74% rejected in Hogencamp, and within the 1% to 5% rejected in Phillips. Thus, in this case, the direct subsidies cannot create a substantial interest.7
Thus, tax dollars that might flow to the county from the city must be considered. While the record could be clearer, the data present leads to the conclusion that if any tax money flows between the jurisdictions, it is much more likely that the flow is from the county to the city, and not the other way around. If this conclusion is wrong, and there is a flow from city to county, it must be so small as to be insignificant relative to the entire operating budget of the county school system.
The total budget (for the year considered by the district court) was $15,891,697; of this, only $2,680,307 came from local taxes. Seventy percent of the budget came from federal and state funds, while the total input from local taxes represented 17.5%. Because most local tax dollars are apportioned and returned to the locality from which they came, the record reveals that only a small portion of this latter percentage might have flowed from the city.
1. County-wide Taxes.
The bulk of the tax money collected throughout the county that is designated for education is divided between the county and city school systems on a pro rata basis based on the number of students enrolled in each school system. For the year considered by the district court, approximately 79% of each tax dollar was allocated to the county (which had 5714 students) while the remaining 21% went to the city (which had 1568 students).8
The defendants, however, presented no evidence that the city paid a disproportionate amount of the ad valorem and excise taxes collected in Escambia County.9 The *781lack of evidence is in contrast to the Phillips v. Andress case, where the defendants proved that the city did in fact pay more in property taxes than was returned to its own school system; nevertheless, the court in Phillips ruled that the monetary difference was insignificant relative to the school budget. 634 F.2d at 951.
This lack of evidence, however, is analogous to the claims of the Phillips defendants that the city in that case paid a disproportionate amount of a county sales tax. The defendants in Phillips presented no evidence attempting to distinguish the county and city residents’ shares of the sales tax. In rejecting this contention of a substantial financial contribution, the Phillips court emphasized the lack of evidence on the question. 634 F.2d at 951. In this case, while claiming that city tax dollars flow to the county, the defendants have not shown us how.10
Yet, even if the defendants here had presented evidence of disproportionate payment by the city, the amount of “overpayment” would still not rise to a significant level. Assuming, arguendo, that the defendants could prove that the city residents (17% of the total population) paid over one-third (34%) of the total county tax receipts, the amount of “overpayment” flowing into the coffers of the county school system would still only be about $130,000, which is less than 1% of the county budget, and thus less than the percentages ruled insignificant in Phillips and Hogencamp. Thus, even if there is some disproportionate tax payment by the city, it is highly unlikely to rise to the level of a substantial interest.
2. Oil Production Tax.
Over half of the local tax dollars distributed to schools are collected under the state oil production tax. Of the $2.6 million of local taxes distributed to the county schools, $1.7 million is from the oil tax. No revenues from this tax are generated from oil production within the city limits of Brewton. Plaintiffs’ Exhibit A at 2. Nevertheless, the defendants attempt to claim some of that money for the city. Their argument is that city residents may own some of the oil, and thus “city” money is used to pay the tax.11
This argument, however, ignores the simple fact that the Alabama oil tax is assessed “at the mouth of the well,” Ala. Code § 40-20-1(1) (1975), and “at the time of severance,” id. § 40-20-3(a). In fact, the tax is to be paid “by the person in charge of the production operations” prior to making payments to the ultimate owners of the oil. Id. Thus, I would conclude that this is “county” money, and therefore this money cannot be said to flow from the city to the county.12
3. Overall Flow of Tax Dollars.
Because (1) the defendants have presented no evidence showing a flow of tax dol*782lars from the city to the county, (2) any possible flow would be insignificant, and (3) the oil money does not flow from city to county, there is no financial support of the county schools by the city, and therefore no substantial interest by the city residents in the county schools.13 And, because I read the Creel, Phillips, and Hogencamp opinions to consider the finances heavily, I am bolstered in my conclusion that the voting scheme in Escambia County is improper.
D. Detrimental Impact of Voting in Escambia County.
Taken together, the minimal student crossover, the equality of bargaining power about shared programs, and the non-existence of financial input from city to county, lead to the conclusion that there is no substantial interest under the Creel, Phillips, and Hogencamp standard. In assessing this minimal interest, this court must also consider the harmful effect of the city enfranchisement on the county electoral process. While the harmful effect is admittedly not as obvious as in the Phillips and Hogencamp cases, the record of the case reveals the harm, both potential and actual. I would reject a suggestion (not made by the majority) that, before we find this type of voting scheme unconstitutional, there must be a city resident on the county board or the city residents must clearly have decided an election. The harmful effect of the voting scheme can be evident long before either of those events occur, and the evidence shows that in this case there has already been impermissible influence in the elections and the operation of the schools.
1. Numerical Harm.
While the final vote totals of the elections considered by the district court do not indicate that the elections would have been different had the city voters not participated, the numbers do reveal the potentially decisive role the city voters play in the elections. This voting power in turn creates the non-numerical harm discussed in the next section.
Of the 14 contested elections considered by the lower court, 7 would have had different winners had the city voted more heavily for the loser. This is most clearly seen in the 1972 Democratic primary runoff, in which the victor, Mr. Beasley, carried the county by only 15 votes (out of almost 6,000); had the city residents preferred Beasley’s opponent by 16 or more votes, the victor would have been a candidate who did not carry the county vote. As it turned out, Beasley took almost 60% of the city vote, and won with what appeared to be a comfortable margin.
*783The city and the county clearly have divergent concerns and preferences. In the 1976 Democratic primary runoff, for instance, the city voters supported Mr. Salter by 72%, while only 55% of the county supported him. It is not implausible that the city voters would prefer one candidate and the county voters another.
Even when the city and county voters tend to agree about the relative rankings of candidates, the city votes can still effect the election results by creating an artificial momentum or an appearance of a solid electoral lead. In the 1976 Democratic primary, for instance, Mr. Salter had a less than 200 vote edge over his closest opponent in the three way contest, in terms of county votes; however, when the city votes were included, Mr. Salter’s edge increased to over 500 votes. The larger edge may well have contributed to Mr. Salter’s victory in the runoff that followed. Another election, the Democratic primary of 1972, illustrated by the following chart, demonstrates this problem:
Candidate City Vote County Vote Total Vote
Beasley 52.4% 37.7% 41.0%
Johnson 28.9% 32.7% 31.9%
Hoomes 18.7% 29.5% 27.1%
TOTAL 100% 100% 100%
Plaintiffs’ Exhibit 6 at 2. If only the county vote was counted, Beasley and Johnson would have been very close going into the runoff. However, because the city vote went so heavily for Beasley, the final vote totals showed Beasley with an almost 10 point lead, instead of the 5 point lead he held in the county. In the actual runoff, Beasley took the county by only 15 votes. I cannot say with any confidence that Beasley would have won without the boost of the city votes in the original primary.
While much of this is admittedly speculative, the data in the record does show that the city could easily control the outcome of an election. This potential creates and contributes to the non-numerical impact of the city votes on the county school board.
2. Non-numerical Harm.
The most obvious harm that cannot be easily quantified is the “chill” over the election and the fact that some candidates may choose not to run for the county school board because they know the city voters participate. A candidate espousing, for instance, the consolidation of all of the county summer programs into one program at a location remote from the city of Brew-ton may face strong opposition within the city, and may thus avoid entering the contest. Even if the city voters have not in the past clearly controlled an election, they still do make up about 25% of the votes cast in the county school board elections. This 25% may well deter potential candidates from participating.
As important as the effect on the actual outcome of the elections is the effect on the actions of the members of the school board once they are in office. Members of the board, of course, are at least to some extent politicians. It is clear from the record that these members run for re-election. It is not difficult to assume that while in office, the members of the school board would prefer to avoid alienating 25% of the electorate. Because of this, the schocl board members (like many politicians) may take their constituents’ views as guides to their actions. A problem arises when, as in this case, their electoral constituency is not exactly the same group that they serve (in this case, the residents served by the county school system).
These intangible harms can still be very real. Given the minimal interest of the city residents in the county schools, the actual and potential harms to the electoral process are higher than this court should be willing to accept.
III. CONCLUSION.
In sum, I would conclude that the City of Brewton does not have a substantial interest in the operation of the Escambia County school system.14 The residents of Brew-*784ton have their own school system, their own school board, and their own vote to elect the council members who appoint the board. Their school board operates independently of, and at arms length with, the county school board. The county school system does not rely on city funds to operate. Because of all of this, the city residents should not dilute the vote of the county residents in electing the county board.15
Thus, respectfully, I dissent.
. In its discussion of the Creel case, the majority perpetuates two statistical errors made in that Fifth Circuit opinion. Majority Op. at 772. First, the Creel opinion emphasizes that almost half of the students in the city schools lived in the county. 531 F.2d at 288. The opinion is silent, however, about how many students in the county system lived in the cities. In attempting to assess the interest of city residents in a county school system, the first figure (found in the Creel opinion) seems almost irrelevant, while the second (not mentioned in Creel) seems crucial. In the Creel opinion, no evidence suggests that any city residents attended the county schools; lacking that information, it is difficult to discern how the Creel court decided that the city residents had a substantial interest in the county school system.
The second error evident in the Creel opinion involves tax dollars that the Creel court concluded flowed from the City of Jasper to the county. The opinion focuses on the $35,501 of Jasper property taxes that went into a pool of money to be divided between the county and the city *776schools. Id. The court noted that of the $394,-524 of taxes collected, 73% went to the county. Thus, according to the Creel opinion, 73% (or $25,915) of Jasper’s $35,501 of tax dollars flowed to the county. Yet, these figures ignore the fact that 27% of county tax dollars flowed to the city. Of the total tax money collected, $106,-521 flowed to the two cities of Jasper and Carbon Hill. Thus, because Carbon Hill was about one-quarter the size of Jasper, it is likely that the Jasper residents paid $35,501 in property taxes, but received far more than that back from the pool of $394,524 collected county-wide. The Creel court, nevertheless, concluded that tax money flowed from Jasper to the county. Id. at 289.
The incorrect statistical inferences found in the Creel opinion, however, do not undercut the validity and correctness of the legal standard announced therein. Thus, although the factual analysis in Creel is flawed, the majority in this case is correct to rely on its legal standard.
. The vast majority of facts presented in this opinion, are drawn from the defendants' responses to requests for admissions and interrogatories propounded by the plaintiffs. These documents were admitted into evidence as exhibits at the hearing before the district court. Record, Vol. 2, at 22-23. At no point in this case has there been any question as to the accuracy of the defendants’ answers. Some of the facts discussed herein are drawn from uncontroverted testimony presented to the district court. Again, there has been no suggestion that this evidence is suspect. In fact, the only real issue before this court is the interpretation of the undisputed facts presented to the district court.
. The essentially personal nature of the interest involved here is highlighted by the defendants’ arguments before the district court, where they argued that "there are 44 people ... going from the city schools out into the county. I submit that that is not insignificant to any of them.” Record, Vol. 2, at 11. I would respond that those individuals have made their own voluntary choice to attend the county schools, and that choice does not give rise to a city-wide interest in voting in the county elections.
. In the present situation, the city voters can absolutely control one side of the joint venture contract (through the city school board), and they can partially control the other side of the same contract (through their electoral influence on the county board). The county residents have only partial control over one side of the agreement.
. This may in fact explain why the city has decided that it does not care to be involved in running the center.
. As with the open door policy, the effect of the majority opinion will be to discourage this type of cooperation between school districts.
. The district court did include a finding of fact stating that the county school board "has received donations in excess of $4,370,000 over the past ten years from sources within the City of Brewton." Dist.Ct.Op. at 8. The record reveals that all of these donations came from a single charitable trust in the city. Plaintiffs’ Exhibit B at 6.
Apart from the relatively small annual average the donations represent, I contend that the donations of a single private trust in the City of Brewton are absolutely irrelevant to the question of whether the entire city has an interest in, and can vote to elect, the county school board. If that single donor desires to influence the county schools, the trust can attach conditions to the donated money.
. Thus, if in fact any money flows from the city to the county, it is limited to the difference in the amount of tax paid in the city and the amount of the tax returned to the city (in this case, about 21%).
. For example, there is no evidence in the record to suggest that the residents of Brewton drink more beer than the county residents, and thus pay a disproportionate share of the county beer tax. But, even if it were proved that the Brewton citizens drank every drop of beer in Escambia County, the tax dollars that would flow to the county would represent less than one-half of one percent of the county schools’ budget, and would thus not be sufficient to create a substantial interest.
. In light of this, the defendants’ response to the following interrogatory posed by the plaintiffs is perhaps not surprising:
12. State whether or not you contend there is a net outflow of funds from residents of the City of Brewton to the Escambia County Board of Education and/or any of the schools under its jurisdiction. If so, state what percentage of the total budget of the County Board of Education and/or the schools under its jurisdiction is derived from residents of the City of Brewton. State how you arrived at this percentage.
Plaintiffs’ Exhibit B at 4. The defendants responded by ”object[ing] to this interrogatory on the grounds that it is vague and ambiguous.” Plaintiffs’ Exhibit B at 8 (answer).
. Again, the defendants have presented no evidence establishing that any of the oil is owned by city residents. In their brief before this court, they can only contend that a "great deal of valuable property located outside the city limits may be owned by city residents who pay the taxes.” Appellees’ Brief at 43 (emphasis added). While it is likely that in fact some city residents do own property in the county, mere likelihoods cannot replace evidence.
. In light of this conclusion, in fact, I suggest that the over $460,000 of oil tax money that is given to the city school system is money that "flows” from the county to the city. Thus, contrary to the arguments of defendants, it seems that the county residents might have an interest in the city schools, rather than the other way around.
. The defendants do propound two additional factors that they suggest support a finding of substantial interest. Neither of these factors, however, are at all relevant.
In oral argument, counsel for the defendants reiterated the district court's finding of fact that 57 out of 350 teachers in the county system reside in the city. Dist.Ct.Op. at 5. This fact, however, is absolutely irrelevant to the question of a city-wide substantial interest in the county schools. A governmental body can certainly require its employees to live within its jurisdiction. See McCarthy v. Philadelphia Civil Service Commission, 424 U.S. 645, 646-47, 96 S.Ct. 1154, 1155, 47 L.Ed.2d 366 (1976). The fact that the Escambia County school board has not done so, and that some teachers have chosen to live outside the school district, does not give rise to an interest by the city in the county schools. As with the 44 crossover students, these 57 teachers have "voted with their feet.”
Of even less relevance is the fact that a substantial percentage of the employees who work within the City of Brewton were educated in the county school system. The defendants admit that this "quality of life” argument is, to say the least, novel. Appellees’ Brief at 43. In the modern world of commuters and mass transit, neighboring communities frequently have cross boundary impact on one another; this, however, does not create an electoral interest in the other community. By the defendants’ logic, the citizens of the county should be allowed to vote in the Brewton city council elections because many county residents commute into Brewton and are thus affected by the actions of the city government. Furthermore, residents of New York City should be permitted to vote in the school board elections of the New Jersey communities across the Hudson River. It should be evident that a generalized interest in the operation and quality of a neighboring community’s government cannot give rise to the substantial interest required by the Creel line of cases.
. A careful analysis of the Creel opinion supports this conclusion. In Creel, 531 F.2d at 289, the Fifth Circuit distinguished a Fourth Circuit case raising similar issues, Locklear v. North *784Carolina State Board of Elections, 514 F.2d 1152 (4th Cir.1975). In Locklear, the Fourth Circuit found this type of challenged voting scheme unconstitutional; Locklear utilized an analysis similar to that found in Creel.
The Creel opinion distinguished Locklear on a number of grounds: (1) the joint programs in Locklear were operated by contractual agreement, (2) the city in Locklear did not finance the vocational center (as happened in Creel), and (3) in Locklear there was no outflow of funds from the city to the county. By those factors, however, this case is almost identical to Locklear. As in Locklear, (1) the joint programs are run by agreement, (2) the county, and not the city, financed the vocational center, and (3) there is no outflow of money to the county. Thus, like Locklear, the voting scheme should be found unconstitutional.
. While the mandate of this court is to enforce the Constitution, and not to implement simple notions of fairness, it still seems relevant to point out that if any group should be voting in the other group's elections, it is county residents who should vote in the city elections. A much larger number of county residents attend city schools than vice versa. The county contributed over $300,000 to a vocational center that the city uses. And a large amount of county money from the oil tax flows each year to the city. If the facts of this case were reversed, I would think that under the Creel standard, the county residents might have a substantial interest in the city schools.