MEMORANDUM OF DECISION

BLUMENFELD, Senior District Judge.

I.

Petitioner was indicted for murder in the first degree on December 6, 1971 by an 18-member grand jury in Fairfield County. He was subsequently tried before a petit jury which returned a verdict of guilty. After exhausting his state remedies, petitioner filed this application for a writ of habeas corpus raising various constitutional claims.

II.

First, both parties agree that, at least prior to 1972, grand juries in Fairfield County were selected from the electorate by the County Sheriff. Sheriff John P. Previdi, County Sheriff at the time of petitioner's indictment, maintained a list of 138 persons from which he made his selections. His predecessor had maintained a similar list. These lists were composed largely of the sheriffs' friends and acquaintances and their friends and acquaintances. In essence, both Sheriff Previdi and his predecessor had unfettered discretion in deciding on the membership of each grand jury.

Second, there is no dispute as to Sheriff Previdi's personal contact with Puerto Ricans. As the trial court found, he personally knew no Puerto Ricans whom he could ask to serve on the grand jury. He did not know the extent of the Puerto Rican electorate nor did he make any attempts to ascertain the neighborhoods in which the Puerto Ricans were concentrated. He was also unaware of the demographic shifts in Fairfield County during his tenure as sheriff.

Third, the parties do not dispute that Puerto Ricans constituted a small but identifiable class in the county. Census figures identified the Puerto Rican population as 1.8% of the total Fairfield County population, and other figures establish that they constituted, at the relevant time, approximately .93% of the electorate eligible for service on the grand jury.

Fourth, both parties agree as to the accuracy of the numbers and statistics involved in this case. Between 1963 and 1971, 738 grand jurors were chosen from the sheriffs' lists and 41 18-person grand juries were empaneled. Of the 738 grand jurors who were selected during this period only two were Puerto Rican. Thirty-nine grand juries had no Puerto Ricans, and two had one Puerto Rican juror each.

III.

While petitioner has no constitutional right to a grand jury array which exactly mirrors the ethnic composition of society, Swain v. Alabama, 380 U.S. 202, 208, 85 S.Ct. 824, 829, 13 L.Ed.2d 759 (1965), the fourteenth amendment does protect him from having the members of his ethnic group substantially underrepresented in the grand jury array. Castaneda v. Partida, 430 U.S. 482, 493, 97 S.Ct. 1272, 1279, 51 L.Ed.2d 498 (1977). As the Supreme Court has indicated in Castaneda v. Partida, supra:

"While the earlier cases involved absolute exclusion of an identifiable group, later cases established the principle that substantial underrepresentation of the group constitutes a constitutional violation as well, if it results from purposeful discrimination."

"[I]n order to show that an equal protection violation has occurred in the context of grand jury selection, the defendant must show that the procedure employed resulted in substantial underrepresentation of his race or of the identifiable group to which he belongs. The first step is to establish that the group is one that is a recognizable, distinct class, singled out for different treatment under the laws, as written or as applied. Hernandez v. Texas, 347 U.S., [475] at 478-479 [ 74 S.Ct. 667, at 670-671, 98 L.Ed. 866]. Next, the degree of underrepresentation must be proved, by comparing the proportion of the group in the total population to the proportion called to serve as grand jurors, over a significant period of time. Id., at 480 [ 74 S.Ct. at 671]. See Norris v. Alabama, 294 U.S. 587 [ 55 S.Ct. 579, 79 L.Ed. 1074] (1935). This method of proof, sometimes called the `rule of exclusion,' has been held to be available as a method of proving discrimination in jury selection against a delineated class. Hernandez v. Texas, 347 U.S., at 480 [ 74 S.Ct., at 671]. Finally, as noted above, a selection procedure that is susceptible of abuse or is not racially neutral supports the presumption of discrimination raised by the statistical showing. Washington v. Davis, 426 U.S., [229] at 241 [ 96 S.Ct., 2040, at 2048, 48 L.Ed.2d 597]; Alexander v. Louisiana, 405 U.S., [625] at 630 [ 92 S.Ct., 1221, at 1225, 31 L.Ed.2d 536]."

A.

No dispute is raised as to either the first or third criterion announced in Castaneda. As to the first criterion, Puerto Ricans clearly constitute "a recognizable, distinct class" which, as the statistics reveal, has been "singled out for different treatment under the laws, . . . as applied." The trial court ruling on petitioner's earlier motion to quash the indictment reached this same conclusion, as did the Connecticut Supreme Court. As noted above, the State does not take a contrary position.

With respect to the third criterion, it is also well-established that the procedure used to choose the grand jurors was a "procedure . . . susceptible of abuse." Fairfield County's procedure afforded its sheriffs even more discretion than did the key-man system in Castaneda, which the Supreme Court considered "highly subjective" and "susceptible of abuse as applied." Id. at 497, 97 S.Ct. at 1281, 1282. Moreover, the Second Circuit has had occasion to warn of the danger implicit in the very selection system here under challenge. It noted:

"Although in this instance there was nothing in the case to show that, in the selection of the grand jury, there was any systematic exclusion of an identifiable group or that the jury list was otherwise not a representative cross-section of the community, the method of selection resting as it does entirely in the hands of one person, the sheriff of the county, leaves much to be desired."

B.

Under Castaneda, the second of the three criteria involves measuring "the degree of underrepresentation" by "comparing the proportion of the group in the total population to the proportion called to serve as grand jurors, over a significant period of time." Castaneda, supra, 430 U.S., at 494, 97 S.Ct., at 1280. In order to make out a prima facie case, petitioner must establish not only that his group was underrepresented, but also that the degree of underrepresentation was substantial. Id. at 493-94, 97 S.Ct., at 1279-1280.

The justification for the "substantial" requirement stems from the need to show intentional conduct. Duren v. Missouri, 439 U.S. 357, 368 n. 26, 99 S.Ct. 664, 670 n. 26, 58 L.Ed.2d 579. Recent decisions of the Supreme Court have indicated that disparate impact alone will not suffice to establish a violation of the fourteenth amendment. Rather, a party claiming that he has been denied equal protection of the laws must establish that there has been an intentional act of discrimination. Washington v. Davis, 426 U.S. 229, 239, 96 S.Ct. 2040, 2047, 48 L.Ed.2d 597 (1976); Village of Arlington Heights v. Metropolitan Housing Development Corp., 429 U.S. 252, 265, 97 S.Ct. 555, 563, 50 L.Ed.2d 450 (1977); Personnel Administrator of Massachusetts v. Feeney, 442 U.S. 256, 272, 99 S.Ct. 2282, 2292, 60 L.Ed.2d 870 (1979).

"Determining whether invidious discriminatory purpose was a motivating factor demands a sensitive inquiry into such circumstantial and direct evidence of intent as may be available." Arlington Heights, supra, 429 U.S., at 266, 97 S.Ct., at 564. Whether the proof offered is sufficient will necessarily vary on a case-by-case basis. Under some circumstances, statistics alone can establish such a clear pattern of discrimination that they cannot be explained on any legitimate grounds. When this occurs, the statistics may amount to circumstantial evidence sufficient to satisfy the intent requirement. Id. at 266, 97 S.Ct. at 564.

How clear this pattern must be appears to vary with the nature of the case. In Arlington Heights, the court indicated that a pattern of discrimination must generally be quite "stark" in order to justify a finding of intent. Id. at 266, 97 S.Ct. at 564. However, as the Court also noted, this requirement is somewhat relaxed in jury selection cases. Id. at 266 n. 13, 97 S.Ct. at 564 n. 13. In Castaneda, supra, a post- Arlington Heights jury selection case, the Court concluded that a showing of "substantial" underrepresentation coupled with an opportunity to discriminate was sufficient to satisfy the intent requirement of the fourteenth amendment. Castaneda, 430 U.S. at 494-95, 97 S.Ct. at 1280-1281.

Consequently, the only issue in dispute on this second criterion is whether the statistics, as agreed to by both parties, are sufficient to establish that Puerto Ricans were "substantially underrepresented" in the grand jury array. Petitioner insists that they are, while the State argues that they are not. In resolving this question it is useful to bear in mind that, in light of the above discussion, "substantial" should be read as "substantial enough to justify an inference of impermissible discriminatory motives."

At its core, the dispute between the parties turns on a difference of opinion over the appropriate way to analyze the statistics. That there should be such a disagreement is not at all surprising given the state of the law on this question. Since courts have begun to consider jury selection cases early in this century they have shown a general trend toward increased sophistication in their use of statistics. Nonetheless, their methodology has varied substantially even within the last decade and a half.

Essentially four different forms of analysis have been used. One of the oldest methods is sometimes called the absolute difference test. D. Baldus J. Cole, Statistical Proof of Discrimination 145 (1980) (hereinafter Baldus Cole). This test was exemplified in the case of Swain v. Alabama, supra, where the Court found that blacks made up 26% of those eligible for jury duty but composed only 10-15% of the venire. The Court subtracted 10-15% from 26% and concluded that the resulting 11-16% disparity did not amount to substantial underrepresentation.

A second approach, perhaps best characterized as the "ratio" approach, Baldus Cole at 145, would lead to a very different conclusion under the same facts. This approach focuses on the percentage of eligible jurors who are excluded. Thus, in Swain, where only 10-15% of the eligible 26% were selected for jury panels, there were roughly 50% fewer blacks than would have been statistically expected. Put another way, any eligible white had more than twice as much chance of becoming a juror as did any eligible black. Arguably this amounted to a 50% exclusion of blacks, a rather substantial degree of underrepresentation. See, e.g., Carmical v. Craven, 457 F.2d 582, 585 (9th Cir. 1971), cert. denied, 409 U.S. 929, 93 S.Ct. 227, 34 L.Ed.2d 186 (1972); Quadra v. Superior Court of San Francisco, 403 F. Supp. 486, 495 n. 9 (N.D.Cal. 1975). Cf. Chance v. Board of Examiners, 458 F.2d 1167, 1171-72 (2d Cir. 1972) (ratio approach applied in challenge to employment test).

The third approach moves away from an examination of percentages and focuses on the differences caused by underrepresentation in each jury. For example, suppose that in Swain the entire array consisted of 120 jurors. Suppose also that from this array ten juries of 12 jurors each were selected. If, in fact, the jurors were drawn at random one would expect that 31 black jurors (26% of 120) would be included in the array and approximately three blacks would show up on each jury. In Swain, however, the actual pool included only 10-15% black members or, under these assumptions, approximately 16 blacks (13% of 120). Thus, each jury would have had an average of slightly more than one-and-a-half black jurors. Some courts have concentrated on the difference between the three jurors expected on each jury and the one-and-a-half jurors actually present. Often they have reasoned that an underrepresentation of only one-and-a-half blacks out of 12 jurors is minimal and that, therefore, the plaintiff has failed to establish substantial underrepresentation. See, e.g., United States v. Kleifgen, 557 F.2d 1293, 1297 (9th Cir. 1977); United States v. Goff, 509 F.2d 825, 826-27 (5th Cir.), cert. denied, 423 U.S. 857, 96 S.Ct. 109, 46 L.Ed.2d 83 (1975); United States v. Jenkins, 496 F.2d 57, 65 (2d Cir. 1974), cert. denied, 420 U.S. 925, 95 S.Ct. 1119, 43 L.Ed.2d 394 (1975) (construing "substantial" in the context of 28 U.S.C. § 1863).

Each of these three tests, however, is inadequate as evidence of intent. Common sense suggests that an absolute percentage difference of 10% can be evidence of intent under some circumstances and not under others. Thus, if the total population consisted of 50 blacks and 50 whites and only ten jurors were chosen, one would expect that the final panel would contain five blacks and five whites. However, if the actual panel contained six blacks and four whites or four blacks and six whites (a 10% absolute difference), one would be unlikely to conclude that this constituted evidence of discrimination. In contrast, assume that a total population contained 10,000 persons, 9,000 of whom were white and 1,000 of whom were black, and that a panel of 1,000 jurors was chosen. If under those circumstances no blacks at all were included in the array, again only a 10% absolute difference, a court could quite reasonable find an intent to discriminate on the basis of race. Similarly, examples which reveal potential errors in the ratio approach can be easily imagined.

Focusing on the number of jurors rather than on percentages does not alleviate the problem either. Where the venire contains numerically fewer black jurors than would statistically be expected, the evidentiary force of the disparity will vary with the total number of black jurors initially expected. Where the number expected is quite small, even slight numerical variations may be more significant than larger disparities would be where blacks are expected in greater numbers. In short, the problem with a test which focuses on the actual number of jurors is that rather than testing for intent it seems to be better designed to test for harm to the defendant. It is designed to show how much difference the underrepresentation will make to the particular complaining defendant rather than to demonstrate and test the intent of the prosecuting State. As Rose v. Mitchell, 443 U.S. 545, 551-59, 99 S.Ct. 2993, 2997-3002, 61 L.Ed.2d 739 (1979), makes clear, harm to the particular criminal defendant is not the relevant consideration.

Following Swain v. Alabama, the Harvard Law Review published an article by Michael O. Finkelstein which, for the first time, developed a rigorous statistical approach to jury selection cases. Finkelstein, The Application of Statistical Decision Theory to the Jury Discrimination Cases, 80 Harv.L.Rev. 338 (1966). Finkelstein's suggested approach was first referred to as "interesting" by the Supreme Court in Whitus v. Georgia, 385 U.S. 545, 552 n. 2, 87 S.Ct. 643, 647 n. 2, 17 L.Ed.2d 599 (1967), then expressly "noted" in Alexander v. Louisiana, 405 U.S. 625, 630 n. 9, 92 S.Ct. 1221, 1225 n. 9, 31 L.Ed.2d 536 (1972), and later fully embraced in Castaneda, supra, 430 U.S., at 496 n. 17, 97 S.Ct., at 1281 n. 17.

Unlike the three previous tests, Finkelstein's statistical decision theory (SDT) provides a mathematically accurate means of testing the probability that a certain degree of underrepresentation could have been the result of random choice rather than intentional discrimination. Where racial factors have not entered into the selection process, one would expect to discover a random distribution of racial representation. If, based on statistics, it appears unlikely that a particular pattern of racial distribution would have resulted from random choice, it is reasonable to infer that discriminatory racial factors have entered into the selection process. See Castaneda, supra, at 494 n. 13, 97 S.Ct. at 1280 n. 13. Unlike the other tests, SDT takes the size of the group into account as one of the factors in the calculations. Thus, the SDT affords a far more accurate measure of intent than do the three other tests discussed above. Accordingly, primary reliance is placed upon it here.

Numerous opinions in which other courts have found no showing of substantial underrepresentation have been brought to the court's attention. As the State indicates in its brief, "Cases such as [these] rely primarily on Swain v. Alabama," which, as noted above, applied an "absolute difference" test. Were such a test to be used here, this court too might be led to reject petitioner's contentions. The total Puerto Rican population was 1.8% of Fairfield County. Two out of 738 grand jurors were Puerto Rican. This amounts to .3% of the array. The absolute difference of 1.5% (1.8%-.3%) cannot be said to be "substantial," as the Court in Swain considered that term. Moreover, the inclusion of even one Puerto Rican grand juror on each of the 39 grand juries previously without Puerto Ricans would grossly overrepresent the actual proportion of Puerto Ricans in the community.

However, if SDT were applied to the facts in this case, even using the premises advanced by the State, it would demonstrate the extreme unlikelihood that this particular distribution of Puerto Ricans could have resulted from racially neutral decisions. The chance of drawing at random no more than two Puerto Ricans in 738 grand juror selections is only 3.28%. The odds of randomly selecting 41 grand juries no more than two of which contained Puerto Ricans is less than four in 100. Viewed from the other side, there was a 96-97% chance that more than two of the 41 grand juries would have had Puerto Rican jurors and that more than two Puerto Ricans would have been selected in 738 selections.

These results are considered by statisticians to be inconsistent with the hypothesis of random selection; if the selection process genuinely had been free from racial considerations, it is extremely likely, even under the State's assumptions, that many more Puerto Ricans would have been selected. Moreover, if any of the initial premises which have been assumed in favor of the State are considered in a light more favorable to the petitioner, the chance that these grand juries were picked without reference to racial factors falls off precipitously.

As noted above, the "substantial underrepresentation" requirement in Castaneda is designed to support the inference that the State acted with an intent to discriminate. Since SDT is the best statistical measure of "substantial underrepresentation," this court finds the petitioner's analysis persuasive notwithstanding the State's authorities to the contrary. When considered in conjunction with the opportunity for abuse inherent in Fairfield County's discretionary selection system, the application of SDT to these facts raises a rebuttable presumption of intentional discrimination.

Indeed, even if these statistics alone were insufficient to establish a prima facie case, two additional factors militate in favor of such a finding. In his own testimony Sheriff Previdi indicated that he had attempted to match Puerto Rican suspects with grand juries containing Puerto Rican members. This suggests that he was aware of the need to be sensitive to racial factors and of the general lack of Puerto Ricans on his panels. Moreover, Sheriff Previdi also indicated that he chose his juries from among his friends and that he had no Puerto Rican friends. Under the teaching of the plurality opinion in Cassell v. Texas, 339 U.S. 282, 70 S.Ct. 629, 94 L.Ed. 839 (1950), this amounts to a virtual confession:

"Our holding that there was discrimination in the selection of grand jurors in this case, however, is based on another ground. In explaining the fact that no Negroes appeared on this grand-jury list, the commissioner said that they knew none available who qualified; at the same time they said they chose jurymen only from those people with whom they were personally acquainted. . . . When the commissioners were appointed as judicial administrative officials, it was their duty to familiarize themselves fairly with the qualifications of the eligible jurors of the county without regard to race and color. The did not do so here, and the result has been racial discrimination."

In short, this court concludes that petitioner has adduced more than enough evidence to establish a prima facie case of purposeful exclusion.

IV.

"Once the defendant has shown substantial underrepresentation of his group, he has made out a prima facie case of discriminatory purpose, and the burden then shifts to the State to rebut that case." Castaneda, supra, 430 U.S., at 495, 97 S.Ct., at 1280; see Rose v. Mitchell, supra, 443 U.S., at 565, 99 S.Ct., at 3005. At best, the State's efforts at rebuttal can be characterized as meager. Sheriff Previdi indicated that on two occasions he had specifically sought out Puerto Ricans for the panel. One such occasion occurred only after the petitioner's motion to quash his indictment was filed and is, therefore, not relevant here. The other time he sought a Puerto Rican, Sheriff Previdi was looking to match a Puerto Rican defendant with a Puerto Rican grand juror. Far from evidence of innocent intent, this act itself constituted a conscious effort to establish a panel based on racial factors, a practice as impermissible as systematic exclusion. Cassell v. Texas, supra, 339 U.S., at 287, 70 S.Ct., at 631 (plurality opinion); Ross v. Wyrick, 581 F.2d 172, 175 (8th Cir. 1978); Harris v. Stephens, 361 F.2d 888, 891 (8th Cir. 1966), cert. denied 386 U.S. 964, 87 S.Ct. 1040, 18 L.Ed.2d 113 (1967).

The State also points out that Sheriff Previdi asked one of his two Puerto Rican grand jurors to sit "several" times but that the grand juror declined. The record indicates, however, that after these refusals the sheriff did not make any efforts to find more cooperative Puerto Ricans. Citations to United States ex rel. Chestnut v. Criminal Court of New York, 442 F.2d 611 (2d Cir.), cert. denied, 404 U.S. 856, 92 S.Ct. 111, 30 L.Ed.2d 98 (1971) and United States ex rel. Epton v. Nenna, 318 F. Supp. 899 (S.D.N.Y. 1970), aff'd 446 F.2d 363 (2d Cir. 1971) are clearly inapposite.

Finally, the sheriff insists that he did not discriminate. While his own declaration is entitled to some weight, a simple protestation that racial considerations played no part in the selection process is not sufficient to rebut a prima facie showing. Castaneda, supra, 430 U.S., at 498 n. 19, 97 S.Ct., at 1282 n. 19; Alexander v. Louisiana, supra, 405 U.S., at 632, 92 S.Ct., at 1226; Turner v. Fouche, supra, 396 U.S., at 361, 90 S.Ct., at 540. Thus, the petitioner has established a prima facie case which the State has failed to rebut, and he is therefore entitled to relief.

V.

"[W]here sufficient proof of discrimination in violation of the Fourteenth Amendment has been made out and not rebutted, this Court uniformly has required that the conviction be set aside and the indictment returned by the unconstitutionally constituted grand jury be quashed."

Notwithstanding this recent expression by the Supreme Court, the State still argues that it would be inappropriate to grant petitioner habeas relief. As the State points out, the petitioner cannot show that the unconstitutional indictment caused him any harm since he was subsequently convicted by a petit jury which was free from constitutional defect. This very argument, however, was rejected in Rose v. Mitchell, supra. In Rose, the Court made it abundantly clear that the injury suffered when an indictment is returned by a racially tainted grand jury is all the injury a petitioner need show in order to quash the indictment and set aside a subsequent conviction. A later finding by a petit jury that a defendant is guilty beyond a reasonable doubt will not render a prior improper indictment harmless error.

The State argues that Rose v. Mitchell is distinguishable because Fairfield County has already changed its system for selecting grand juries. Here, unlike Rose, there is an affidavit from the sheriff indicating that the old, unfettered-discretion system has been replaced by a new system designed to insure random selection. Thus, the State concludes that there is no need to grant habeas relief in order to deter future constitutional violations by the county.

The State's argument appears to be predicated on the assumption that the underlying justification for the result in Rose v. Mitchell was the concept of deterrence. A careful reading of the opinion, however, does not support that conclusion. The Court in Rose was concerned with the "appearance of justice" and the "integrity of the judicial process." Id. 443 U.S. at 555-56, 99 S.Ct. at 2999-3000. To allow the petitioner to remain incarcerated where the trial itself was set in motion by a body tainted with racial bias would be to condone such behavior and implicitly to wink at "odious" and "pernicious" racial discrimination. Id. at 555, 99 S.Ct. at 2999. It is because racial discrimination in the selection of a jury "strikes at the fundamental values of our judicial system and our society as a whole" that it is inappropriate to treat it as harmless error after a valid conviction. Id. at 556, 99 S.Ct. at 3000.

VI.

SO ORDERED.