This action challenging the validity of the 1970 reapportionment of Chicago’s aldermanic wards on the grounds of racial and ethnic gerrymandering is here again on appeal from the judgment following the second trial. Many of the pertinent facts (including the 1970 map of Chicago’s wards) are set out in our earlier opinion (reversing and remanding the case for further proceedings) and will not be repeated here. Cousins v. City Council of City of Chicago, 466 F.2d 830 (7th Cir. 1972), cert. denied 409 U.S. 893, 93 S.Ct. 85, 34 L.Ed.2d 151.
*914On retrial the district court found that the evidence fails to establish that the boundaries of any of the wards except the 7th are the product of purposeful or invidious racial or ethnic discrimination. The court found that defendants intended to change what they believed was a black majority in the 1961 7th ward into an overwhelmingly white majority. To correct the 7th ward situation, the court deemed it sufficient to redraw the boundary between the 7th and 8th wards so that each had a majority of black residents of approximately 52%, and ordered a special election November 27, 1973 in the modified 7th ward. Taxable costs were awarded to plaintiffs, but all other relief sought by them was denied. The special election was held and the person elected is serving.
Plaintiffs appealed from the judgment, except for specified portions including those with respect to the 7th and 8th wards, and they appealed from a supplemental order denying them attorney’s fees. Defendants cross appealed from described portions of the judgment, including those affecting the 7th and 8th wards.
At the second trial much of the evidence from the first trial was reintroduced. The most important additions were the second deposition of Alderman Keane and a sample ward map, offered by plaintiffs and referred to as the Singer map. Mr. Rubin Singer prepared this map using the figures available when the subcommittee had drawn the 1970 map, achieving equality of population within the same narrow limits as the subcommittee, maintaining contiguity, and making the wards as compact, i. e., as nearly square, as possible. He did not take into account the location of black or Spanish-language people.
In our earlier opinion, we stated that although no minority racial or ethnic group is entitled to have any particular voting strength reflected in the city council, “such strength must not be purposefully minimized on account of . race or ethnic origin.” 466 F.2d at 843. The district court was satisfied that, except for the 7th ward, the proof did not establish purposeful minimization. In examining whether its findings were clearly erroneous (mindful of our duty of specially close scrutiny, 466 F.2d at 837), we shall first consider the overall result of 15 black majority wards in the 1970 map, and any evidence that this number instead of a greater one was the result of a purpose to minimize black voting strength. We shall then examine the evidence with respect to the choice of ward lines in particular regions pointed out by plaintiffs and considered by the district court. Finally we will focus on the ward boundaries claimed to be gerrymandered to the detriment of Puerto Ricans and the Spanish-language group.
I. The possibility of drawing the inference of purposeful minimization from the overall result.
In our earlier opinion, we indicated that plaintiffs had fallen short of producing sufficient persuasive circumstantial proof that the ward boundaries were the result of purposeful discrimination.1 We suggested that a demonstration that one or more maps could be drawn, “following an objective standard, rationally related to redistricting and indifferent to race” producing greater voting strength for minority groups, would be persuasive. 466 F.2d at 843. The only map introduced pursuant to this suggestion was the Singer map. Defendants note that the mapmaker is a “parti*915san,”2 but do not otherwise make any claim that the map is not objective or that it is defective.
In their arguments before the district court on remand and before us, the parties utilized the Singer map primarily to aid in their analysis of particular regions. At this point, we shall inquire whether the 1970 districting map, when evaluated in comparison with a racially neutral districting plan shows such difference as to be itself a basis (or perhaps one basis in conjunction with others) for inferring a purposeful minimization of the voting strength of the racial minority.
A table is set out in the margin comparing the number of wards created by the Singer map and the 1970 map with given percentages of black population.3 At the outset we note as argued by plaintiffs that the Singer map created 16 black majority wards (out of a total of 50 wards) while the 1970 map 4 produces only 15 black majority wards. Looking further, and assuming that there is probably enough political strength in a 35% minority to be significant for our present purpose, we have compared the number • of wards with 35% or more black population under the 1970 map and the Singer map. There are 19 such wards under the Singer map and 18 under the 1970 map.
The record contains considerable evidence that Alderman Keane was aware of the racial compositions of various areas of the city, and that the choices of some particular ward lines were motivated by that consideration.5 It is difficult to say that the differences between the number of wards with certain percentages of black residents in the 1970 map and in the Singer map are sufficient to be highly persuasive that there was purposeful minimization of black voting strength in the 1970 map, and even in the light of the evidence of awareness of race and consideration thereof by Keane, it is our judgment that the finding of the district court is not clearly erroneous.
II. Evidence as to specific boundaries.
On remand, the parties and trial court focused on the extent to which racial considerations might have affected ward boundaries in particular regions of the city. We now review the results of this approach.6
*916A. The 7th and 8th wards.
Defendants cross-appeal from the district court’s order modifying the boundaries of the 7th and 8th wards to eliminate invidious discrimination.7
At the outset the controversy on appeal concerning the 7th ward is not moot even though the election has been held in the modified 7th ward. If there were to be no redistricting in 1974, the present ward boundaries would apply in the elections in 1975 and 1979. Although the city council has adopted a resolution directing the preparation of a new redistricting ordinance to govern the 1975 election,8 a new ordinance has not yet been enacted. Plaintiffs also represent that the subcommittee designated to prepare the new districting has requested a city department to prepare a new plan based upon the shape and location of wards in the 1970 ordinance.
In reviewing the district court finding of purposeful discrimination in drawing the 7th ward boundaries, it is convenient to consider two distinct stages of the formulation of the 7th ward as constituted in the 1970 ordinance: first the changes in formulation leading to the final Bell-student map, and, second, the changes which occurred during the subcommittee proceedings.
1. Comparison of the 1961 map with the final Bell-student map.
A rectangular appendage of the L-shaped 9th ward occupied the southeastern corner of Chicago on the 1961 map. The main portion of ward 9 extended north from the southern city limit and west of ward 10; the appendage extended eastward to the Indiana state line and lay south of the 10th ward. Student intern Taylor’s uncontroverted testimony is that the eastern appendage of the 9th ward was eliminated in the drafting of the Bell-student map in order to achieve compactness. The new 10th ward, occupying the southeastern corner of the city, was a rectangle, with greater north-south than east-west dimensions.9
The changes in the 9th and 10th wards were made before the 7th ward was drawn. As a result of adding the eastern appendage of the 9th ward to the 10th, the 10th ward would be overpopulated if its 1961 northern boundary were maintained.10 To compensate, the northern boundary of the 10th was moved south.
The southern boundary of the 7th was and is the northern boundary of the 10th, and drawing that line further south necessitated either a southerly shift of the 7th, or creation of a new ward between the 7th and 10th. On the 1961 map, the northern boundary of the 7th was at 67th street. Taylor’s undisputed testimony is that when first drawn by Bell and the students, the northern boundary of the 7th was south of 73rd street, thereby excluding the residence of Alderman Bohling (a Caucasian). Bell and the students regularly consulted a list of the names and ad*917dresses of all incumbents and attempted not to draw an incumbent out of the ward most substantially corresponding to his 1961 ward. Apparently for this purpose, the northern boundary of the 7th was shifted north to 71st street, thereby including Bohling’s residence. Finally, in order to approach the desired numerical result, and because of the limitation of Lake Michigan at the east, the western boundary was shifted eastward, thereby decreasing the population in the 7th ward.
Comparing the racial composition of the 7th ward on the 1961 map and of the 7th ward on the final Bell-student map,11 using 1970 census figures, it appears that the percentage of black residents in the 1961 7th was 48.3% and in the Bell-student 7th only 29.4%. The parties had believed the percentage of black residents in the 1961 7th ward was higher, and stipulated it was 55%-60%. Subsequent to the first appeal it was announced that black persons had been un-dercounted in the 1970 census, to the extent of 7.7% nationwide, although the distribution of the undercount in particular areas has not been determined. Presumably the percentages we have derived from census figures should be increased to some unknown extent.
Thus the 7th ward on the 1961 map contained a majority or very near majority of black persons, while they are approximately a 30% minority in the 7th ward of the final Bell-student map. Of course changes in other ward boundaries occurred as a result of this change. Given the evidence that the choices of boundaries were responses to legitmate or nonjusticiable concerns11b such as creating a more compact 9th ward, drawing the 7th so as to include an incumbent alderman, and seeking equality of population, and the absence of evidence pointing specifically to any racial motivation,12 the difference between the percentages of black persons in these two areas will not, in our opinion, support an inference that the change in ward boundaries was purposeful minimi2;ation of black voting strength.13 We are aided in reaching this conclusion by the failure of the Singer map to produce more black majority or substantial black minority wards in the southeastern part of Chicago, encompassing the wards altered by the students, than did the 1970 map.13b
*918(ii) From the final Bell-student map to the 1970 map.
We now look for support for the inference of purposeful minimization of black voting strength in the subcommittee proceedings. The procedure followed is set out in our earlier opinion. 466 F.2d at 834-835. In forming the 7th ward, Alderman Keane began at its southern boundary, a natural line formed by the Chicago Skyway on the southwest and the Calumet River on the southeast. This boundary between the 7th and 10th differed from the student map only in that tract 4610 was included in the 7th ward as Alderman Keane drew it, and not in the 10th ward as the students drew it. 4610 is a triangular tract which, when put in the 7th, filled in the natural Skyway-Calumet River boundary of that ward. Its population is 22.55% black.
Plaintiffs’ contention with regard to diminution as a result of the subcommittee proceedings focuses on the allocation of two groups of census tracts between the predominantly black 8th ward and the predominantly white 7th ward, specifically tracts 4309, 4311, and 4312, tracts with 85-90% black residents, located next to the northwest portion of the 7th ward (as enacted in 1970), and tracts 4604 and 4603, tracts with high percentages of white residents, located in the central western portion of the 7th ward as enacted.
Two of the three black tracts and both white tracts were within the 7th ward when first called off by Alderman Keane. By the end of the first subcommittee session on the 7th ward, both white tracts were within the 8th ward and all three black tracts were in the 7th ward.14
At the end of the next session concerning these wards, one of the predominantly white tracts had been shifted back into the 7th and the two predominantly black tracts had been transferred into the 8th.15
During the third revision, considerably later, the predominantly white tract under consideration and remaining in the 8th ward was transferred to the 7th and the predominantly black tract under consideration and remaining in the 7th was transferred to the 8th.16
The transfers of two predominantly white tracts into the 7th ward and three predominantly black tracts into the 8th ward would be consistent with a purpose of minimizing black voting strength in the 7th ward by adding black residents to the 8th, where black residents already had a high majority. At least awareness of the racial composition was shown by Alderman Keane’s comment to Alderman Cousins, the black alderman representing the 8th ward, concerning the transfer of tract 4603 from the 8th to the 7th — “I understand that there are a lot of Polish people in this Tract 4603 and that you don’t want to represent them.” Any tendency of these facts to support a finding that the changes were motivated by a purpose to minimize black voting strength is offset by the evident greater compactness of the 7th and 8th wards as finally drawn when contrasted to the earlier subcommittee versions.17
*919Moreover the net changes subsequent to the final Bell-student map had only the effect of decreasing the percentage of black residents in the 7th ward from 29.4% to 26.9% or somewhat larger figures if the census undercount of black persons be considered.
Summarizing our review of the district court finding o’f purposeful discrimination in drawing the 1970 7th ward, we consider: (1) the fact that although the change from the 1961 7th ward to the 1970 7th ward represented, in a sense, the loss of one black majority ward, 'the 1970 map produced 15 black majority wards (as compared with 11 on the 1961 map), with little persuasive proof that the number was held at 15 with a racially discriminatory purpose; (2) the fact that the greatest changes in the ward and its racial composition occurred in the preparation of the Bell-student map and were responses to legitimate or nonjusticiable concerns; (3) the fact that although there was evidence consistent with an inference that some of the changes made by the subcommittee were racially motivated, there were also legitimate explanations for some of them in terms of compactness, and in any event, a relatively small reduction (2.5%) in the size of the approximately 30% black minority is insignificant as related to the issues here.18 On the entire evidence we are left with the definite and firm conviction that a mistake has been committed,19 and conclude that the finding as to the 7th ward is clearly erroneous. We understand that defendants do not seek a new election nor other change in representation of the 7th ward during the current term, which has less than one year to run. Therefore, we limit our reversal so as not to affect the tenure of the aider-man now serving during the current term.
B. The nth and 16th wards.
Plaintiffs assert that the findings of the district court with regard to the 14th and 16th wards entitle plaintiffs to relief as a matter of law. Plaintiffs’ theory is that as first drawn by Bell and the students, the wards in.which white aldermen Burke and Sheridan reside were majority black; that pursuant to Alderman Keane’s explicitly racial instructions, the early student map was redrafted to place incumbents Burke and Sheridan in majority white wards; and that the changes involved in such incorporation created one or more additional such wards and one or more fewer wards of majority black population.
A careful examination of the findings reveals no basis for plaintiffs’ theory. Properly, and consistently with plaintiffs’ theory, the district judge found that Alderman Keane made racially explicit statements to Bell and the students ;20 that pursuant to Keane’s instructions the Bell-student map was revised so as to place the residences of white aldermen Sheridan and Burke in white majority wards; 21 and that the *920motivation for these changes was to ensure the reelection of these two “bright young men” by providing them with substantially white electorates.
But the district court made a further finding: the revisions made to “accommodate” the white incumbents did not, in drawing white majority wards which would include those aider-men, produce a smaller number of black majority wards than had existed before the revision. Because the question raised by this action is whether the defendants purposefully minimized black voting strength and is not merely whether defendants’ line-drawing was affected by racial considerations, this critical finding, if not clearly erroneous, is fatal to plaintiffs’ claim. After carefully scrutinizing the record, we conclude that the finding that the changes in this region did not minimize black voting strength is not clearly erroneous.
The 14th, 15th, and 16th wards are located in the southwest part of the city and form a region roughly approximating a square in shape. The population in the southeast corner of this region is almost entirely black; this area is adjoined at the north and west by an inverted L-shaped region virtually all the residents of which are white. Because of the clearly delineated racial makeup, it is clear that the region occupied by these three wards could not be divided into compact wards, of which two have black majorities. A region roughly similar to the region occupied by the 14th, 15th, and 16th wards on the 1970 map was also divided into three wards on the 1961 map and again a roughly similar region was so divided on the final Bell-student map. Each map produced two substantially white majority wards and one almost completely black ward to the southeast. Furthermore, the Singer map does not significantly alter the proportions in the same general region; instead, it creates a number of virtually all white wards and one solidly black ward.22
The record does not contain a facsimile of the first draft of the Bell-student map which purportedly created two black majority wards in nearly enough the same region to include the residences of Burke and Sheridan. Instead plaintiffs rely upon Alderman Keane’s comment that the wards were “too black” (the inference is that he would not make such a statement unless two black majority wards were created by the Bell-student draft), upon one intern’s interpretation of Keane’s remarks as meaning the wards were majority black as drawn; and upon Taylor’s testimony that subsequent to the conversation with Keane the wards in this region were shifted some unspecified distance to the west thereby incorporating more whites into the western portions of wards 14 and 15 on the 1970 map.23 Although this evidence has some probative force, we cannot say that the district judge’s finding that no substantial diminution in black voting strength resulted was clearly erroneous. Accordingly, notwithstanding the fact that the remand gave them an opportunity to develop additional statistical evidence on this issue, plaintiffs did not satisfy their burden of proof with regard to these wards.
*921III. Dilution of voting strength of Spanish-language groups.
We shall first consider plaintiffs’ claims with regard to the group delineated as Puerto Rican; then, we shall examine plaintiffs’ contentions about the Spanish-language group.
On appeal, plaintiffs do not press their initial argument that looking at the Puerto Rican population of Chicago on a city-wide basis, it appears that the voting strength of Puerto Ricans has been purposefully diminished. The district court found that there are only 49,500 persons in the Puerto Rican category; that they are scattered over an area consisting of three or four wards; and that it would be impossible to create a Puerto Rican majority in any single ward. We conclude that these findings are not clearly erroneous.24
The only regional challenge with regard to the Puerto Rican group involves changes between the 1961 and 1970 boundaries of the 26th and 31st wards. Specifically, plaintiffs emphasize the exclusion from the 31st ward of the Puerto Rican region to the east (census tracts 2425 and part of 2426) and the incorporation of the non-minority tracts in the west (census tracts 2306 and 2313).
There is ample support in the record for the district court’s finding that in accord with Alderman Keane’s wishes, Bell, in drawing the final Bell-student map, extended the 31st ward as far west as possible in order to avoid the inclusion of additional Puerto Ricans in Keane’s ward. See 466 F.2d at 838-839. The boundaries on the Bell-student map and on the 1970 map are very similar. Relief would be justified therefore had plaintiffs been able to demonstrate a cognizable dilution of Puerto Rican voting strength.
The district court found, however, that the failure to include the Puerto Rican regions in the 31st ward caused no discernible loss of Puerto Rican voting strength; we conclude that this finding is not clearly erroneous. The 1970 census figures reveal that as enacted ward 31 had a Puerto Rican population of approximately 25%. If the boundaries of the 1961 map had been retained ward 31 could have been as much as about 32% Puerto Rican.25 We do not think this change is sufficiently substantial, especially in light of the corollary decrease in Puerto Rican voting strength in the 26th ward if the 1961 boundaries are retained for the 31st ward.26
We also conclude that the district court’s finding that there was no significant diminution in black voting strength as a result of the attempt to keep the 31st ward to the west is not clearly erroneous. Had the 1961 boundaries of ward 31 been retained, the black population would have increased only from approximately 1.4% to 1.7%.
During the second trial plaintiffs attempted to broaden the minority group being considered from that of Puerto Ricans to the Spanish-language group. We think the district court acted within its discretion in concluding that plaintiffs’ failure to amend their complaint to conform with this additional proof and to add a named plaintiff representing the Spanish-language *922group prevented consideration on the merits.
Furthermore, as noted by the district court, the evidence introduced indicates that the Spanish-language group totals only about 80,000 persons, contains an unknown number of aliens who cannot vote in elections, and is so scattered as to constitute at best a substantial minority. Under these circumstances, we are not persuaded by the fact that the Singer map created one ward with approximately 42% Spanish-language population while under the 1970 map the most substantial Spanish-language minority comprises only 33% of the population of the ward.
IV. Compactness and burden of proof.
In our earlier opinion we indicated that deviations from compactness may be relevant in establishing invidious discrimination but that otherwise lack of compactness presents state law questions. 466 F.2d at 833-34. Plaintiffs have not demonstrated that the district court failed to follow this principle.27
Plaintiffs argue that the district court applied incorrect rules governing the burden of proof. Specifically plaintiffs object to the failure of the district court to shift to defendants the burden of establishing the absence of discrimination following plaintiffs’ introduction of historical evidence and statistical evidence and to the district court’s requirement that plaintiffs’ burden of persuasion be measured by the “clear and convincing” standard.
Turning to plaintiffs’ first argument, it does not appear to be addressed to the strictly evidentiary question of which party has the burden of coming forward with evidence28 but to the more substantive inquiry about whether the district court should have applied the “strict scrutiny” test, placing the burden of justifying the 1970 ordinance on the defendants. As urged by plaintiffs,
“[P]revious decisions have indicated, strict scrutiny means that the State’s system is not entitled to the usual presumption of validity, that the State rather than the complainants must carry a ‘heavy burden of justification,’ that the State must demonstrate that its [legislation] has been structured with ‘precision,’ and is ‘tailored’ narrowly to serve legitimate objectives and that it has selected the ‘less drastic means’ for effectuating its objectives. . . . ” San Antonio School District v. Rodriguez, 411 U.S. 1, 16-17, 93 S.Ct. 1278, 1288, 36 L.Ed.2d 16 (1973).
Because most of the Supreme Court decisions touching upon the allocation of burdens in racial gerrymandering cases involve challenges to multimember districts, we first consider the bearing of such authority upon actions where single-member districts are claimed to dilute minority voting strength. The Supreme Court has long recognized that there is a “tendency” of multimember districts “to submerge minorities” but has not held that multimember districts are illegal per se. See Whitcomb v. Chav-is, 403 U.S. 124, 142-143, 159, 91 S.Ct. 1858, 29 L.Ed.2d 363 (1971) and cases cited therein. We think that if the cases reveal an unwillingness of the Supreme Court to shift the “burden of justification” (as urged by plaintiffs here) in causes involving the somewhat “suspect” multimember districts, such an allocation would be even less likely in the context of a challenge to single-member districts.
Although we recognize that fundamental constitutional concerns about voting *923and racial discrimination are intertwined in racial gerrymandering cases, plaintiffs have not cited nor are we aware of any such action, involving either multimember or single-member dis-tricting, in which the “strict scrutiny” analysis, as defined in San Antonio, has been applied.
In Whitcomb, an action challenging state statutes creating a multimember district for the election of state senators and representatives, the Supreme Court stated that the “challenger [must] carry the burden of proving that multi-mem-ber districts unconstitutionally operate to dilute or cancel the voting strength of racial . . . elements.” The Court reversed the three-judge district court, which had invalidated the reapportionment plan, in part because the Court did not think that plaintiffs had adequately demonstrated an actual dilution of voting strength.
In Whitcomb, unlike our case, plaintiffs conceded that the multimember dis-tricting was not purposefully designed to dilute the vote of minorities although such minimization was alleged to result. Whitcomb, supra, at 149, 91 S.Ct. 1858. Furthermore, when distinguishing the facts before it, from actions involving allegedly purposeful devices, the Supreme Court stated;
“It needs no emphasis here that the Civil War Amendments were designed to protect the civil rights of Negroes- and that the courts have been vigilant in scrutinizing schemes allegedly conceived or operated as purposeful devices to further racial discrimination. There has been no hesitation in striking down those contrivances that can fairly be said to infringe on Fourteenth Amendment rights.” Whitcomb at 149, 91 S.Ct. at 1872.
Nonetheless, the above language does not mandate the “strict scrutiny” test urged by plaintiffs. In the racial gerrymander case cited with approval following this passage, the three-judge district court did not place the burden of justifying the legitimacy of the multi-representative districts upon the state. Sims v. Baggett, 247 F.Supp. 96, 107-110 (M. D.Ala., 1965). The other racial gerrymander case cited, Gomillion v. Lightfoot, 364 U.S. 339, 81 S.Ct. 125, 5 L.Ed. 2d 110 (1960), is not significant here as it involved a ruling on a motion to dismiss; all allegations of plaintiffs were assumed to be true.
More recently, the Supreme Court has affirmed the judgment of a three-judge district court which allocated the burden to the plaintiffs to establish that a Texas multi-member constituency apportionment scheme minimized voting strength of black people and Mexican Americans. See White v. Regester, 412 U.S. 755, 93 S.Ct. 2332, 37 L.Ed.2d 314 (1973), aff’g Graves v. Barnes, 343 F.Supp. 704, 720 (W.D.Tex. 1972). The Supreme Court did not review the trial court’s disposition of the challenge to a single-member districting scheme in a county, but we note that once again the burden had been placed squarely upon the challengers and no burden of justification was placed unon the state. Graves, 343 F.Supp. at 735. The Supreme Court cases of Kil-garlin v. Hill, 386 U.S. 120, 87 S.Ct. 820, 17 L.Ed.2d 771 (1967) and Wright v. Rockefeller, 376 U.S. 52, 84 S.Ct. 603, 11 L.Ed.2d 512 (1964) also indicate that the “strict scrutiny” test has not been applied in racial gerrymandering cases.29
*924In sum, it appears that in Sims, Kilgcurlin, and Wright (to the extent the Supreme Court cited them with approval or affirmed), the trial courts made their determinations whether plaintiffs had established unconstitutional racial gerrymandering upon all the evidence, without allocating the burden of justification to the defendants. See generally Burns v. Richardson, 384 U.S. 73, 88-89, 86 S.Ct. 1286, 16 L.Ed.2d 376 (1966). We conclude that the district court in the instant case properly followed this approach.
We note that plaintiffs rely in their argument upon the facts suggesting past discrimination referred to in our earlier opinion, 466 F.2d at 840-841. These facts were in the record on the first trial. It is implicit in our opinion that we did not deem them adequate to satisfy plaintiffs’ burden. Indeed we considered whether plaintiffs had shown enough so that the matter could be resolved on a limited remand by the production of a map like the Singer map. We rejected this solution, and decided that a new trial of the issues was appropriate. 466 F.2d at 843-844.
Plaintiffs’ final argument is that the challenger’s burden of persuasion should be measured by the more lenient “preponderance of the evidence” standard rather than the “clear and convincing” standard applied by the district court here. There is language in Sims, Kilgarlin and Wright indicating that the presumption of constitutionality adheres to reapportionment statutes even when challenged on the ground of racial gerrymandering.30 On the other hand in Graves, in a portion of the opinion not reviewed by the Supreme Court, the district court appears to have required only preponderating evidence of substantial if not egregious discriminatory effects caused by reapportionment. Graves, 343 F.Supp. at 735. We need not resolve this question here. From our analysis of the evidence and the observations of the district court, we are confident that the district court’s assumption that a preponderance of the evidence would have 'been insufficient unless the proof could be found clear and convincing made no difference in the finding made.
V. Attorney’s fees and other costs.
Plaintiffs’ application for attorney’s fees for over 1,000 hours of services and other non-taxable costs was denied by the district court. This court joins with the district court in acknowledging that attorneys for the plaintiffs have been diligent. Not only have they caused a searching inquiry into an important public issue, but they have compiled a record which is full of important information which may be helpful in testing future districting for invidious discrimination. Their exposure of official lack of candor concerning the Bell-student map, 466 F.2d at 835, was itself substantial public service.
Nonetheless, particularly in view of the ultimate result, we find no abuse of discretion in denial of their application for fees and other expenses.
Defendants argue that the award of taxable costs to plaintiffs was erroneous and must be reversed if we reverse the judgment as to the 7th and 8th wards. Defendants, however, failed to appeal from the portion of the judgment awarding costs.
Insofar as the judgments and orders appealed from ordered a change in the *925boundaries of the 7th and 8th wards of the City of Chicago, they are reversed, except that said reversal shall not affect the status of the aldermen of those wards, now serving, before the end of the current term of office. In all other respects they are affirmed.
. Purposeful minimization of the voting strength of a minority racial or ethnic group is the sort of conduct proof of which is unlikely to be direct or specific. A finding that the challenged districting reflects such discrimination must usually rest on inferences from circumstances. One approach, where sucli a claim is made is to compare the number of black majority wards in the challenged map with another map which might equally reasonably be drawn, but with more such wards, and assess whether the difference is more probably explained by a purpose to discriminate than by legitimate considerations.
. Singer was a member of the Board of Directors of plaintiff Independent Voters of Illinois at the time this action was instituted.
. The figures on racial breakdowns were supplied by the Census Bureau subsequent to the drafting of the 1970 ordinance. The figures in brackets represent the number of wards in each category after the change ordered by the district court below.
Percent black population Number of wards created with such percentage Singer map 1970map
0-15% 27 28
15-25 1 1
25-35 3 3 [2]
35-45 2 3
45-50 1 0
50-55 2 0 [2]
55-65 2 1
65-75 1 1
75-85 1 2 [2]
85-100 10 11_
50 50 [50]
. The comparisons are made with the 1970 map as enacted, and not as changed by the district court.
. Most of the direct evidence of racial awareness and consideration was introduced at tlie first trial and is detailed in our first opinion. 466 F.2d at 830. Some is noted in our discussion of plaintiffs’ regional challenges, infra.
. As a practical matter, it is difficult to determine the extent to which choices of boundaries within the region are the result of prior legitimate changes in boundaries outside the region, and the effect, if any, of changes within the region upon other subsequently drawn wards outside the region.
. The plan adopted by the district court results in both the 7th and 8th wards having a majority of black residents, and, as it happens, would produce a total of 16 wards with such majority, as would the Singer map. Although the new 7th ward appears reasonably compact, the new 8tli ward appears to deviate from maximum compactness to a greater extent than did the 8th ward on the 1970 map.
. The resolution refers to a portion of our opinion with respect to the population disparities which first became apparent after the 1971 election. 466 F.2d at 844.
. The wards on the Bell-student map and in proceedings before the subcommittee were designated by letters rather than numbers. It is convenient, however, to refer to them by the number ultimately given to the ward containing most substantially the same area. This number is the same as the most substantially corresponding 1961 ward. The 1970 ward with a particular number almost always (and as a result of design) contains the residence of the alderman who represented the 1961 ward with the same number. As the changes in the map occur, the various wards are usually referred to as if there were a continuity in particular wards.
. Part of census tract 5501 and all 5502 were added to the 10th.
. The record does not reveal the exact boundaries of the ward as first drawn; therefore comparison is made only with the final form.
b. Following Gaffney v. Cummings, 412 U.S. 735, 751-754, 93 S.Ct. 2321, 37 L.Ed.2d 298 (1973), it is not clear which redistricting decisions designed to achieve political ends or to allocate political power are nonjusticiable. The Supreme Court has stated, however, that the “fact that ‘district boundaries may have been drawn in a way that minimizes the number of contests between present incumbents does not in and of itself establish invidiousness.’ Burns v. Richardson, 384 U.S. 73, 89 n. 16 [86 S.Ct. 1286, 1295, 16 L.Ed.2d 376] (1966). Cf. Gaffney v. Cummings, [412 U.S.) at 752 [93 S.Ct. 2321, 37 L.Ed.2d 298]; Taylor v. McKeithen, 407 U.S. 191 [92 S.Ct.1980, 32 L.Ed.2d 648] (1972).” White v. Weiser, 412 U.S. 783, 797, 93 S.Ct. 2348, 2353, 37 L.Ed.2d 335 (1973).
. There was testimony only that Bell was aware of the general racial effects of the change in shape of the 7th ward.
. The district court concluded that the 7th and 8th wards were drawn with the invidious purpose of minimizing black voting strength. In part the finding appears to be based on the difference in percentages of black residents between the 1961 7th and the student map 7th above noted, and in part on some switching of areas back and forth between the 7th and 8th, hereafter mentioned. The court found an absence of discriminatory purpose in the drafting of the Bell-student map. We agree. Because, however, the only substantial reduction in the percentage of black population between the 1961 and the 1970 7th wards occurred in the drafting of the Bell-student draft, the finding that the 7th ward was intentionally drawn to create a white majority ward instead of a black majority ward, is in conflict with the finding that there was no discriminatory purpose at that stage.
b. The same approximate region is occup-pied by wards 6, 7, 8, 9, 10, 21 and 34, on the 1970 map and wards A, B, D, E, F, J, and K on the Singer map. On either map, there are four wards with a majority of *918black persons (6, 8, 21, and 34; I), E, J, and K), no wards with a substantial minori-tiff of black persons (35% to 50%), and three wards with less than 35% black population (7, 9, 10; A, B, F).
. Specifically, at the first call tracts 4311 and 4312 (predominantly black) were in the 7th ward as were tracts 4603 and 4604 (predominantly white). Tract 4309 (predominantly black) was within the 7th ward. Alderman Keane next transferred 4604 into the 8th and 4309 into the 7th. Finally 4603 was transferred into the 8th. Keane apparently used an incorrect ward designation when describing this last transfer; although the census tracts being called were clearly in the 8t!i ward, designated O at the subcommittee proceedings, he explained he was about to redraw ward 21, designated N by the subcommittee.
. Specifically, tract 4603 was shifted into the 7th and 4311 and 4312 into the 8th.
. 4604 was moved from the 8th to the 7th and 4309 from the 7th to the 8th.
. Tracts 4309, 4311, and 4312, when in the 7th formed a bulge to the west near the northern portion of the 7th, and tracts 4603 *919and 4604, when in the 8th, caused an appendage of that ward to jut into the center of the 7th and leave a relatively narrow waist at one point.
. This action does not present the question whether this degree of reduction would be constitutionally cognizable where a plaintiff alleged that the value of his or her vote within a particular ward had been impaired by invidious racial gerrymandering of the boundaries of that ward. The plaintiffs here are not dissatisfied with the wards in which they live, but have standing because they belong to a group living in many wards whose voting strength is allegedly purposefully diluted. See 466 F.2d at 845.
. See United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 92 L.Ed. 746 (1948).
. See 466 F.2d at 839.
. Sheridan represented the 16tli ward under the 1961 districting plan. By 1970, his old ward had become approximately 80% black. Under the 1970 plan he resides in the 15th ward which is 91% white. (Sheridan ran in 1971 and was defeated.) Burke represented the 14th ward under the 1961 districting plan. By 1970, his old ward was approximately 80% white. Under the 1970 plan he resides within the 14th ward which is 93% white. (He ran and was reelected in 1971.)
. On the Singer map the southeastern corner of the roughly corresponding region is occupied by ward N which is 92.6% black. AVard U, which incorporates part of ward 14 is only .9% black. AVard O, including part of ward 15 is 3.9% black. Ward H, including part of ward 15, the southwestern corner of ward 16, and a western sliver of ward 17, and the most eastern portion of ward 18, is 30% black. The incorporation of ward 18’s black population in ward H, however, leaves ward G, which includes the western portion of ward 18 with only .3% black population. In other words, there is only a shift in location of one 30% black minority ward; under the Singer map it is ward H, under the 1970 map, it is ward 18.
. Of course, without any evidence of the extent of this shift, it is impossible to deduce what the black population was prior to the westward movement.
. We note that plaintiffs’ failure to introduce statistics reflecting the Puerto Rican population of the Singer map wards prevents us from determining whether the Singer map provided wards with more substantial Puerto Rican minorities than the 1970 map.
. Because of the absence of statistics in the record showing the Puerto Rican population of census tracts, this figure was derived by calculating the effect on the new 31st ward Puerto Rican population of deleting census tract 2425 and half of census tract 2426 and adding census tracts 2306 and 2313. Looking at the figures most favorably to plaintiffs, we assumed that the Puerto Rican population of tracts 2306 and 2313 is 0% although the record reveals only that it is below 10%.
. The record enables us to calculate that assuming no other changes, deletion of census tract 2425 and part of census tract 2426 from ward 26 would decrease the Puerto Ri-can population from 20% to 17%.
. We also note that despite plaintiffs’ contention otherwise it appears the district court did consider whether the wards were compact under Illinois law and concluded that the wards were sufficiently compact to satisfy state law. This finding is not clearly erroneous.
. Because plaintiffs and defendants have come forward with extensive documentary and testimonial evidence from which the relevant inferences can be drawn, if plaintiffs’ argument is directed to shifting the burden of coming forward, it is of no merit at this stage of the proceedings.
. In Kilgarlin v. Hill, the Supreme Court in a per curiam opinion affirmed the judgment of the district court insofar “as it held that appellants [plaintiffs] had not proved their allegations that [the state legislative reapportionment act] was a racial . . . gerrymander violating the Fourteenth Amendment [or] that it unconstitutionally deprived Negroes of their franchise.” 386 U.S. at 121, 87 S.Ct. at 821. It is clear from the opinion of the three-judge district court that the burden was allocated to plaintiffs to prove their allegations and that no “burden of justification” was placed on defendants. Kilgarlin v. Martin, 252 F.Supp. 404, 434-441 (S.D.Tex.1966).
In Wright v. Rockefeller, the Supreme Court affirmed, accepting the findings of the majority of the three-judge district court that “appellants [plaintiffs] failed to prove that the New York Legislature was either motivated by racial considerations or in fact drew the districts on racial lines.” 376 U.S. at 56, 84 S.Ct. at 605.
. In Sim.i the district court stated: “Testing the constitutionality of the reapportion- ■ ment Acts of the Legislature of Alabama by the foregoing standards, and after according to each all presumptions in favor of its constitutionality, we have concluded . . . ” 247 F.Supp. at 105. In Kilgarlin the district court indicated that “clear proof” is needed ' to establish that the “Legislature harbored the sinister design of racial discrimination.” 252 F.Supp. at 435. In Wright, the Supreme Court quoted and discussed without disapproval concurring District Judge Feinberg’s statement that “Plaintiffs have a difficult burden to meet in attacking the constitutionality of this state statute.” D.C., 211 F.Supp. 460 at 469; 376 U.S. at 55-57, 84 S.Ct. 603.