**70Article 18 of the Amendments to the Massachusetts Constitution, as amended by arts. 46 and 103 of the Amendments, **71known as the "anti-aid amendment," prohibits in § 2, cl. 2, the "grant, appropriation or use of public money ... for the purpose of founding, maintaining or aiding any church, religious denomination or society." This case presents the question *694whether two grants of public funds to renovate an active church that has been identified as a "historic resource" under the Community Preservation Act (act), G. L. c. 44B, are categorically barred by the anti-aid amendment, or whether the constitutionality of such grants must be evaluated under the three-factor test we have applied under Commonwealth v. School Comm. of Springfield, 382 Mass. 665, 675, 417 N.E.2d 408 (1981) ( Springfield ), to payments made to other private institutions. Also presented is the follow-up question: if the three-factor test applies, do the grants satisfy its requirements?
We conclude that the constitutionality of such grants must be evaluated under our three-factor test: a judge must consider whether a motivating purpose of each grant is to aid the church, whether the grant will have the effect of substantially aiding the church, and whether the grant avoids the risks of the political and economic abuses that prompted the passage of the anti-aid amendment. We also conclude that, in light of the history of the anti-aid amendment, a grant of public funds to an active church warrants careful scrutiny. Because the judge applied this three-factor test incorrectly in denying the plaintiffs' motion for a preliminary injunction to prohibit disbursement of these grants, we vacate the order denying the motion. As to the grant to preserve the stained glass windows in the main church building, we remand the case to the Superior Court for entry of an order allowing the plaintiffs' motion for a preliminary injunction barring disbursement of the grant. As to the grant to fund a "Master Plan" to preserve all three of the buildings belonging to the church, we remand for further proceedings consistent with this opinion.2
Background. The Acton Congregational Church (church), an affiliate of the United Church of Christ, is an active church with **72a congregation of over 800 members. It describes its mission thusly:
"The mission of Acton Congregational Church ... is to preach and teach the good news of the salvation that was secured for us at great cost through the life, death, and resurrection of Jesus. The church encourages each individual to accept the gift of Christ and to respond to God's love by taking part in worship, ministry to one another, and the Christian nurture of people of all ages. With the guidance of the Holy Spirit, we are called as servants of Christ to live our faith in our daily lives and to reach out to people of this community and the world with love, care, and concern for both their physical and spiritual needs."
The church stands in the Acton Centre Historic District (historic district), an area that has served as a center of town life since the establishment of the town of Acton (town) in 1735. The church owns and maintains three adjacent buildings in the historic district: the main church building, the John Fletcher House, and the Abner Hosmer House. The main church building was built in 1846. Today, it is used for worship services and religious educational programs; it also houses a local day care center, meeting spaces for various community *695groups, and a thrift shop. The two houses, also built in the mid-Nineteenth Century, originally were private residences but were later acquired by the church and are now rented to local families.
The town is one of 172 municipalities in Massachusetts that have adopted the act, which establishes a mechanism for funding projects relating to open space, historic resources, and community housing.3 G. L. c. 44B. In 2015, the church submitted two grant applications to the town's Community Preservation Committee (committee), which makes recommendations in accordance with the act to the town meeting regarding "the acquisition, preservation, rehabilitation and restoration of historic **73resources."4 G. L. c. 44B, § 5 (b) (2). See G. L. c. 44B, § 7.
The church's first application was for a $49,500 grant to fund a "Master Plan for Historic Preservation" for all three of its buildings (the Master Plan grant). The church proposed to hire an architectural consultant to develop a plan for their renovation and preservation; the proposed work would include "a thorough assessment of the [c]hurch building envelope, including windows, doors, siding, roof, chimney, bell tower, skylights, and fire escapes." The church noted "[s]pecific areas of concern" for the building, including its bell tower and brass chandelier.
The church's second application was for a $51,237 grant to fund the restoration and preservation of the main church building's stained glass windows, which were installed in 1898 (the stained glass grant). According to the church's application, the "most prominent" of the windows depicts Jesus and a kneeling woman; another window features a cross and the hymnal phrase, "Rock of Ages Cleft for Me." The proposed work would include replacing parts of the glass, sealing the glass, and installing new glazing so that the windows-which currently have a "cloudy" exterior and "cannot be appreciated outside the church"-will be given "complete transparency."
The church explained in its applications that, due to declining membership and contributions, it lacked the funds necessary both to preserve its buildings and to fully serve the needs of its congregation without financial assistance from the town:
"As you may know, mainstream churches have not been growing for years, and the financial strain is significant. [The church] has weathered the storm better than many churches, but the reality is that we have had to cut programs and personnel. The cuts can further exacerbate the financial problem by not offering the congregation what draws them to their church. With that in mind, the long list of maintenance and capital improvement projects get[s] delayed before we cut programs, but there are many things that we've had to fix."
Consistent with the requirements of the act, the committee held a public hearing on *696the church's applications and voted unanimously **74to recommend the two grants. The town approved them both at a town meeting.
The town imposed several conditions on the grants. First, it required that the church convey to the town a "historic preservation restriction" in the buildings that would be "perpetual to the extent permitted by law." Second, it specified that no funds would be disbursed to the church except as reimbursements for specific expenses incurred in connection with the projects, and only after the town could verify, based on submitted invoices, that those expenses were "consistent with the project scope presented" in the church's applications.
The plaintiffs, a group of town taxpayers, commenced this action in the Superior Court under G. L. c. 40, § 53, which permits taxpayers to act "as private attorneys general" to enforce laws designed to prevent abuse of public funds by local governments. LeClair v. Norwell, 430 Mass. 328, 332, 719 N.E.2d 464 (1999). The plaintiffs sought a declaratory judgment that the grants to the church violate the anti-aid amendment, and requested injunctive relief to prevent their disbursement.5
In denying the plaintiffs' motion for a preliminary injunction, the judge relied on the three-factor test we first set forth in Springfield, 382 Mass. at 675, 417 N.E.2d 408. We applied the test in that case to determine whether a statute that authorized the public funding of special education placements of public school students in private schools violated the anti-aid amendment. Id. at 667, 417 N.E.2d 408. The three factors are: "(1) whether the purpose of the challenged statute is to aid private schools; (2) whether the statute does in fact substantially aid such schools; and (3) whether the statute avoids the political and economic abuses which prompted the passage of [the anti-aid amendment]." Id. at 675, 417 N.E.2d 408.6 We cautioned that these factors "are not 'precise limits to the necessary constitutional inquiry,' but are instead guidelines to a proper analysis." Id., quoting Colo v. Treasurer & Receiver Gen., 378 Mass. 550, 558, 392 N.E.2d 1195 (1979). We also recognized that each factor was "interrelated,"
**75and that any conclusion "results from a balancing" of the factors as applied to the facts of each case. Springfield, supra at 675, 417 N.E.2d 408.
The judge here determined that the plaintiffs bore a heavy burden to overcome the presumption of the act's constitutionality because, although the plaintiffs were challenging the constitutionality of the grants to the church, those grants were awarded pursuant to the act. Thus, as to the first factor, the judge determined that she must "examine the purpose of the [act]," and concluded that the purpose of the grants under the act was "to preserve historic resources, and not to aid the [c]hurch[ ]." As to the third factor, the judge found that "[t]here is no credible evidence that the grants under the [act] are economically or politically abusive or unfair," noting that "[t]he application and approval procedures for grants under the *697[act] operate without regard to the applicant's makeup or purpose." The judge concluded that, even if the plaintiffs were to satisfy the second factor, which she was "not convinced they can," they still had "no likelihood of success on the merits" because their failure to satisfy the first and third factors "preclud[ed] them from overcoming the presumption of constitutionality that favors the [act]."
The judge also granted the town's motion for a protective order to stay discovery until thirty days after entry of a decision on the preliminary injunction. The plaintiffs appealed from the denial of their motion for a preliminary injunction and the allowance of the protective order. We granted their application for direct appellate review.
Discussion. In a taxpayer suit such as this, the taxpayers collectively are acting as a private attorney general seeking under G. L. c. 40, § 53, "to enforce laws relating to the expenditure of tax money by the local government." LeClair, 430 Mass. at 332, 719 N.E.2d 464. In order to obtain a preliminary injunction, the plaintiffs must show a likelihood of success on the merits and that the requested relief would be in the public interest; they need not demonstrate irreparable harm. See id. at 331-332, 719 N.E.2d 464.
The plaintiffs claim that the judge made two errors of law in her decision denying their motion for a preliminary injunction. First, they argue that the judge erred by applying the three-factor test articulated in Springfield, contending that this test only applies where the challenged grant of public funds is to aid a private school or institution, and not where the challenged grant is to aid a church. Second, they contend that, even if the three-factor test properly applies to public aid to churches, the judge misapplied **76the test. To rule on these claims of error, we must look first to the history and evolution of the anti-aid amendment.
1. The history and evolution of the anti-aid amendment. Our original Declaration of Rights, adopted in 1780, provided in art. 3 for the direct public support of religion, continuing the Colonial practice of using tax revenues to support the "public Protestant teachers of piety, religion and morality[,]" see Colo, 378 Mass. at 556 n.10, 392 N.E.2d 1195, which essentially meant support of the Congregational Church. See T.J. Curry, The First Freedoms, Church and State in America to the Passage of the First Amendment, 163-164, 174-175 (1986) (Curry); S.E. Morison, A History of the Constitution of Massachusetts 24 & n.1 (1917) (Morison).7
Even before it was mandated by the Declaration of Rights in 1780, the "quasi-religious establishment" of the Congregational Church had provoked heated conflict. Id. at 24. See generally 1 W.G. McLoughlin, New England Dissent 1630-1833, The Baptists and the Separation of Church and State, 547-568 (1971) (McLoughlin). During the American Revolution, Baptists protested the religious assessments *698with acts of civil disobedience; in retaliation, mobs attacked them on the pretext that they were Tories. See Curry, supra at 163. When the Constitution was submitted to the people for ratification, forty-five towns rejected art. 3, most of them because it provided public support to the Congregational Church. See id. at 167-169; McLoughlin, supra at 626-631. After art. 3 was enacted, the Baptists challenged the religious assessments in court, and other denominations followed. See McLoughlin, supra at 636-659.
After decades of "lawsuits, bad feeling, and petty persecution," Morison, supra at 24, the Massachusetts Constitution was amended in 1833 with art. 11 of the Amendments enacted to substitute for art. 3. Article 11 guarantees the equal protection of "all religious sects and denominations"-not just the Christian **77denominations protected under art. 3-and effectively ended religious assessments. The next year, the Legislature enacted a statute providing that "no citizen shall be assessed or liable to pay any tax for the support of public worship ... to any parish or religious society whatever, other than to that of which he is a member." St. 1834, c. 183, § 8. See Morison, supra at 38-39.
But the issue of public support for religious institutions was far from resolved by art. 11. It was raised again in the Constitutional Convention of 1853, which adopted art. 18 of the Amendments to prevent the appropriation of public funds to sectarian schools.8 See 3 Debates and Proceedings in the State Convention 1853, at 613-626 (1853) (Debates of 1853); Morison, supra at 59. The debates from the Convention indicate that art. 18 did not arise in response to any actual funding of sectarian schools in Massachusetts, but from fear of the sectarian conflict that would result if such funding were to occur. See Debates of 1853, supra at 615, 618-620.9
The delegates worried that competing claims from various denominations would quickly deplete public funds for education. In the words of one delegate: "[I]f we take the position that a part of this fund may be given to one denomination, another may come in and claim the same privilege, and another, and another, until the fund is completely exhausted ...." Id. at 620. But the **78delegates were equally fearful of the political controversies that were bound to ensue. See id. at 619, 624. One delegate warned that making public funds available *699to religious institutions would be like throwing "a firebrand into ... town meetings." Id. at 624. The "object" of art. 18, he explained, was "to extinguish [that] firebrand, so that it shall not be possible to rekindle it." Id. Having seen until 1833 how public financial support for churches could provoke such animosity between citizens, the delegates were eager to remove the controversial issue of religion from politics. See id. at 624-625.
In fact, religious tensions were on the rise in 1853, as Massachusetts faced a massive influx of immigrants, most of them driven here from Ireland by the famine caused by a potato blight that devastated the nation's harvest. See generally O. Handlin, Boston's Immigrants, A Study in Acculturation, 25-53 (rev. ed. 1979). In 1841, about 10,000 Irish immigrants arrived in Boston; in 1846, that number had risen to more than 65,000. Id. at 242. By 1850, more than one-fourth of Boston residents were Irish. Id. at 243. Hostility toward Irish Catholics grew among those who felt threatened by the combined forces of mass immigration, urbanization, and industrialization. See Haynes, The Causes of Know-Nothing Success in Massachusetts, 3 Am. Hist. Rev. 67, 70-76 (1897) (Haynes). Rumors spread about a "papal plot" to spread Catholic influence throughout the government and in particular the public school system. See Holt, The Politics of Impatience: The Origins of Know Nothingism, 60 J. Am. Hist. 309, 323-324 (1973). These anti-Catholic sentiments were well known to the framers of art. 18. Indeed, some delegates believed (and historians today agree) that art. 18 was itself targeted specifically against Catholic schools.10 See Debates of 1853, supra at 615-617; J.R. Mulkern, The Know-Nothing Party in Massachusetts, The Rise and Fall of a People's Movement, 42 (1990) (Mulkern); Shapiro, The Conservative Dilemma, The Massachusetts Constitutional Convention of 1953, 33 New Eng. Q. 207, 224 (1960). See also Wirzburger v. Galvin, 412 F.3d 271, 281 (1st Cir. 2005), cert. denied, 546 U.S. 1150, 126 S.Ct. 1165, 163 L.Ed.2d 1128 (2006).
It bears noting that art. 18, along with all the amendments adopted by the 1853 Convention, failed to be ratified by the **79people in 1853. Morison, supra at 63. However, in 1854, the Know-Nothing Party, running on an anti-foreign and in particular an anti-Catholic platform, won a surprising political victory in Massachusetts that secured both the governorship and control of the Legislature. See Haynes, supra at 67-68. Article 18 was revived by the Know-Nothing government, Mulkern, supra at 94, 105-106, and ratified by special election in 1855, Morison, supra at 64.
However, the adoption of art. 18 did not end the controversy over public support for religious institutions. Public dissatisfaction with art. 18 grew when, due to its "rather uncertain language," private religious schools and hospitals continued to receive public funding. Bloom v. School Comm. of Springfield, 376 Mass. 35, 39, 379 N.E.2d 578 (1978). See Loring, A Short Account of the Massachusetts Constitutional Convention 1917-1919, 6 New Eng. Q. 1, 10 (1933). In 1913, the Legislature requested this court's opinion on whether art. 18 "adequately prohibit[ed]" the appropriation of public funds "for maintaining or aiding any church, religious denomination or religious society, or any *700institution, school, society or undertaking which is wholly or in part under sectarian or ecclesiastical control." Opinion of the Justices, 214 Mass. 599, 599-560, 102 N.E. 464 (1913). The Justices were in agreement that art. 18 prohibited appropriations to primary and secondary schools under sectarian control, but not to schools of higher education. Id. at 601, 102 N.E. 464. The Justices were divided, however, on whether art. 18 allowed appropriations to a church or religious denomination; four Justices were "of opinion that such an appropriation is prohibited by the Constitution and its Amendments," while three Justices "incline[d] to the opposite conclusion." Id.
Faced with this uncertainty, delegates to the Constitutional Convention of 1917 sought "to tighten the prohibition of public support for religious education" and "to protect State and municipal treasuries from the growing pressure of interest groups in search of private appropriations." Springfield, 382 Mass. at 673, 417 N.E.2d 408. The result was art. 46 of the Amendments, a substantially revised version of art. 18 that was "sweeping in its terms." Bloom, 376 Mass. at 39, 379 N.E.2d 578. Article 46 broadened the prohibition on the use of public funds to encompass not only private religious schools but all private institutions, whether secular or religious, and, in the last clause of § 2, specifically prohibited the "grant, appropriation or use of public money ... for the purpose of founding, maintaining or aiding any church, religious denomination or society **80."11
By its terms, the revised anti-aid amendment applied to all institutions not under public control. Its proponents recognized that, in the fight over public funds, private institutions of all kinds-whether religious or not-were equally likely to compete. See 1 Debates in the Massachusetts Constitutional Convention, 1917-1918, at 62-70, 163-168 (1919) (Debates of 1917-1918). As one of the amendment's chief supporters explained during the debates: "[I]f you let the bars down everything else will come in." Id. at 118. The decision to appropriate funds to one private institution would lead to "a thousand other [s]" asking for the same. Id. The anti-aid amendment was intended to keep those bars up, protecting public funds from religious and secular institutions alike.12
*701Still, the delegates to the Convention voiced many concerns that were specific to religious institutions, as reflected in the last clause of § 2 of the revised anti-aid amendment. As we have summarized in the past:
"Proponents of [the anti-aid amendment] urged that liberty of **81conscience was infringed whenever a citizen was taxed to support the religious institutions of others; that the churches would benefit in independence and dignity by not relying on governmental support; and, more generally or colloquially, that to promote civic harmony the irritating question of religion should be removed from politics as far as possible, and with it the unseemly and potentially dangerous scramble of religious institutions for public funds in ever-increasing amounts."
Bloom, 376 Mass. at 39, 379 N.E.2d 578, citing Debates of 1917-1918, supra at 68, 74-79, 161-164.
The anti-aid amendment that emerged from the 1917 Convention is the amendment-with some revisions adopted in 1974, not relevant here13 -that applies today. It currently provides:
"No grant, appropriation or use of public money or property or loan of credit shall be made or authorized by the [C]ommonwealth or any political subdivision thereof for the purpose of founding, maintaining or aiding any infirmary, hospital, institution, primary or secondary school, or charitable or religious undertaking which is not publicly owned and under the exclusive control, order and supervision of public officers or public agents authorized by the [C]ommonwealth or federal authority or both, [with exceptions not relevant here]; and no such grant, appropriation or use of public money or property or loan of public credit shall be made or authorized for the purpose of founding, maintaining or aiding any church, religious denomination or society."14
Art. 18, § 2, as amended by arts. 46 and 103.
2. Does the three-factor test in Springfield apply to public aid to churches? Section 2 of the anti-aid amendment contains two clauses: the first clause prohibits the grant of public funds "for the purpose of founding, maintaining or aiding" any institution that is not publicly owned or under exclusive public control, including **82schools and hospitals; the second clause prohibits the grant of public funds "for the purpose of founding, maintaining or aiding any church, religious denomination or society." Art. 18, § 2, as amended by arts. 46 and 103. The plaintiffs contend that the three-factor test in Springfield applies only where the challenged grant of public funds is to a private school or institution under the first clause, and should not be applied where the challenged grant is to an active house of worship under the second clause, as in this case. Rather, the plaintiffs argue that the second clause requires an "unequivocal and unqualified" ban on the grant of public funds to churches. We disagree. *702This is the first time that we have been asked to consider the constitutionality of a grant of public funds to a church under the second clause of the anti-aid amendment. All of our prior decisions under the anti-aid amendment since its revision in 1917 have considered the actual or contemplated grant of public funds or assistance to private schools or institutions under the first clause. See Helmes v. Commonwealth, 406 Mass. 873, 874, 550 N.E.2d 872 (1990) (funding for repair of memorial battleship); Attorney Gen. v. School Comm. of Essex, 387 Mass. 326, 327, 439 N.E.2d 770 (1982) ( Essex ) (transportation for private school students); Springfield, 382 Mass. at 665, 666, 417 N.E.2d 408 (funding for special education programs in private schools); Colo, 378 Mass. at 551, 392 N.E.2d 1195 (payment of legislative chaplains' salaries); Bloom, 376 Mass. at 36, 379 N.E.2d 578 (textbooks for private school students). See also Opinion of the Justices, 401 Mass. 1201, 1202, 514 N.E.2d 353 (1987) (tax deduction for expenditures on tuition, textbooks, and school transportation); Opinion of the Justices, 357 Mass. 846, 847-848, 259 N.E.2d 564 (1970) (vouchers for private school students); Opinion of the Justices, 357 Mass. 836, 837-838, 258 N.E.2d 779 (1970) (reimbursement of private schools for secular educational services).
In Springfield, 382 Mass. at 675, 417 N.E.2d 408, we declared that "there are no simple tests or precise lines by which we can determine the constitutionality" of grants challenged under the first clause of the anti-aid amendment. Instead, we devised the three-part test as "guidelines to a proper analysis," id., quoting Colo, 378 Mass. at 558, 392 N.E.2d 1195, focusing on the purpose of the grant, the extent to which the grant aids the private institution, and whether the grant "avoids the political and economic abuses" that led to the passage of the anti-aid amendment, all of which must be carefully balanced in determining its constitutionality. Springfield, supra at 675, 417 N.E.2d 408.
This rejection of "simple tests [and] precise lines" is equally appropriate when evaluating the constitutionality of a grant of **83public funds under the second clause of the anti-aid amendment. Id. The operative language in each clause is identical: both provide that no "grant, appropriation, or use of public money ... shall be made or authorized" "for the purpose of founding, maintaining or aiding" one of the enumerated private institutions. Art. 18, § 2, as amended by arts. 46 and 103. In both clauses, the specific reference to "purpose" demands an inquiry into both the making of a grant and its purpose.15 Where the language of the two clauses is essentially the same, our interpretive framework is appropriately also the same. See, e.g., Alliance, AFSCME/SEIU, AFL-CIO v. Secretary of Admin., 413 Mass. 377, 384, 597 N.E.2d 1012 (1992) ("Words occurring in different places in the Constitution and its amendments ordinarily should be given the same meaning unless manifestly used in different senses" [citation omitted] ); Opinion of the Justices, 384 Mass. 820, 823, 425 N.E.2d 750 (1981) (interpreting word "items" in §§ 3 and 5 of art. 63 of Amendments to have same meaning).
Moreover, even if we did not look to our interpretation of the first clause for guidance, we could not read the second clause as an absolute ban on grants to churches, *703because the second clause by its own terms calls for a case-by-case analysis. The words of the second clause are not: "No grants shall be made to any church." Rather, the second clause prohibits only grants that are made "for the purpose of founding, maintaining or aiding any church," and we cannot know that every grant to a church will be for that purpose. The categorical prohibition urged by the plaintiffs therefore invites the danger of overbreadth-and of hubris. We do not presume that we have the wisdom or imagination to contemplate every possible grant of public funds to a "church, religious denomination or society" and be certain that all of them, regardless of purpose, effect, or historical context, would be barred by the anti-aid amendment.
A categorical prohibition also invites the risk of infringing on the free exercise of religion, a right guaranteed under the First Amendment to the United States Constitution ("Congress shall make no law respecting an establishment of religion, or prohibiting the free exercise thereof"); art. 2 of the Massachusetts Declaration **84of Rights ("no subject shall be hurt, molested, or restrained, in his person, liberty, or estate, for worshipping God in the manner and season most agreeable to the dictates of his own conscience; or for his religious profession or sentiments; provided he doth not disturb the public peace, or obstruct others in their religious worship"); and the anti-aid amendment itself. See art. 18, § 1, as amended by art. 46 ("No law shall be passed prohibiting the free exercise of religion").
This was the risk addressed in Trinity Lutheran Church of Columbia, Inc. v. Comer, --- U.S. ----, 137 S.Ct. 2012, 2017, 198 L.Ed.2d 551 (2017) ( Trinity Lutheran ), where a church in Missouri was denied a public grant to resurface its playground. In contrast with the Massachusetts anti-aid amendment, the Missouri Constitution imposes a categorical prohibition on any grant of public funds "in aid of any church, sect[,] or denomination of religion."16 Id. As a result, when a church preschool and day care center applied for a grant under a general government program to purchase a new playground surface made from recycled tires, the State's Department of Public Resources rejected its application, based on "a strict and express policy of denying grants to any applicant owned or controlled by a church, sect, or other religious entity." Id. The Supreme Court of the United States held that the department's policy of excluding a church from a government program "solely because it is a church," id. at 2025, "imposes a penalty on the free exercise of religion that must be subjected to the 'most rigorous' scrutiny," id. at 2024, quoting Church of Lukumi Babalu Aye, Inc. v. Hialeah, 508 U.S. 520, 546, 113 S.Ct. 2217, 124 L.Ed.2d 472 (1993).17
*704We do not interpret the Massachusetts anti-aid amendment to **85impose a categorical ban on the grant of public funds to a church "solely because it is a church." Trinity Lutheran, 137 S.Ct. at 2025. Rather, under our three-factor test, whether a church can receive such a grant depends on the grant's purpose, effect, and the risk that its award might trigger the risks that prompted the passage of the anti-aid amendment. Such an analysis would surely not bar the grant of public funds to a church preschool to provide a safer surface for its playground. Cf. Essex, 387 Mass. at 333-334, 439 N.E.2d 770 (State funding to provide transportation to students attending private schools did not violate anti-aid amendment because it was "a general program to help parents get their children, regardless of their religion, safely ... to and from ... schools" [citation omitted] ).18
Therefore, we conclude that the judge did not err in declining to interpret the second clause of the anti-aid amendment as a categorical prohibition on the grant of public funds to churches.
3. Application of the three-factor test. The plaintiffs contend that, even if the constitutionality of the grant should be determined under the three-factor test, the judge erred as a matter of law in her application of that test. We agree, and discern two distinct errors of law.
First, in determining whether the grants at issue would violate the anti-aid amendment, the judge focused primarily on the constitutionality of the act itself rather than on the constitutionality of the award of the two grants at issue.19 Analysis of the act's constitutionality would have been appropriate if the act itself authorized the appropriation of public funds to a church or other private institution within the scope of the anti-aid amendment. See, e.g., Helmes, 406 Mass. at 875, 877-878, 550 N.E.2d 872 (applying three-**86factor test to statute authorizing expenditure of public funds for repair of World War II battleship under control of charitable corporation); Springfield, 382 Mass. at 668, 675-683, 417 N.E.2d 408 (applying three-factor test to statute authorizing school committees to contract with private schools to provide special needs education where public schools could not meet special needs).20 *705Here, however, the act simply establishes a procedure for municipalities to make discretionary grants to projects relating to open space, historic resources, and community housing. See G. L. c. 44B, §§ 5, 7. Nothing in the act itself specifically authorizes the expenditure of funds to assist churches or religious institutions.
For this reason, the constitutionality of the act itself was not challenged by the plaintiffs, and is not at issue in this case. What was challenged, and is at issue, is the constitutionality of specific discretionary grants made pursuant to the act. Therefore, "the familiar principle of statutory construction that affords a statute a presumption of constitutionality validity," Springfield, 382 Mass. at 674, 417 N.E.2d 408, does not apply to the constitutional analysis of these grants, and the judge erred in applying that presumption. The grants themselves enjoy no such presumption of constitutionality.
Second, the judge's focus on the constitutionality of the act rather than of the grants also rendered erroneous her analysis of the first and third factors.21 As to the first factor, the judge relied on the language of the test as it was applied to the statutes at issue in Springfield and Helmes, and therefore considered whether the legislative purpose of the act was to aid churches. The judge instead should have considered whether the primary purpose of the committee in recommending the grants was to aid this particular church rather than to serve the proper purpose of historic preservation.
Accordingly, we now apply the three-factor test to the proposed grants themselves. On this record, we conclude that the plaintiffs **87are likely to succeed on the merits of their claim with respect to the stained glass grant, but that further discovery is needed to evaluate their claim as to the Master Plan grant.
a. Purpose. The first factor to be considered is whether the proposed grants are "for the purpose of founding, maintaining or aiding [a] church." Art. 18, § 2, as amended by arts. 46 and 103. In ascertaining the purpose of a challenged grant, our cases concerning aid to private schools are instructive. In Springfield, 382 Mass. at 678, 417 N.E.2d 408, we upheld the constitutionality of a statute that funded special education programs in private schools for children whose needs could not adequately be met in public schools, finding that its "primary purpose" was "to benefit public schools and individual children." We saw no evidence of any "hidden legislative purpose" to aid the private schools themselves. Id. at 677, 417 N.E.2d 408. See Essex, 387 Mass. at 331, 439 N.E.2d 770 (statute authorizing provision of transportation to private school students held constitutional based on "avowed purpose" to benefit children and lack of any "hidden purpose to maintain private schools"). In contrast, in Bloom, 376 Mass. at 42, 379 N.E.2d 578, we declared unconstitutional a statute requiring public school committees to lend textbooks to children attending private schools because we could infer from this statutory scheme no other purpose than to aid private schools "in carrying out their essential function." We determined that it made no difference under the anti-aid amendment that the textbooks *706were to be lent to the students rather than to the private schools they attended. Id. at 47, 379 N.E.2d 578. What mattered was that the statute made use of public money or property for the purpose of "maintaining or aiding" the private schools. Id. at 42, 379 N.E.2d 578.
Here, historic preservation is the stated purpose of the committee in awarding these grants to the church. That stated purpose is consistent with the town's decision to make the grants contingent on a historic preservation restriction in the three buildings. Such a restriction would limit the church's ability to make changes to the buildings in the future, thereby ensuring that the historic value of those buildings is not diminished over time. Thus, the plaintiffs' burden under the first factor is to demonstrate a "hidden ... purpose" to aid this particular church. Springfield, 382 Mass. at 677, 417 N.E.2d 408.22
We conclude that the record before us is insufficient to determine **88whether such a hidden purpose existed. The plaintiffs here sought to depose a person, to be designated by the town under Mass. R. Civ. P. 30 (b) (6), as appearing in 435 Mass. 1501 (2001), to testify regarding the town's "[c]onsideration and approval of the applications for the [c]hurch [g]rants," and the communications among town officials, employees, and committee members regarding the applications, but the judge denied the plaintiffs this discovery for purposes of the motion for preliminary injunction when she granted the town's motion for a protective order. Where the anti-aid amendment itself focuses on the "purpose" of a grant to a church, and where the first factor to be considered under our test is the purpose of the grant, a plaintiff is entitled to reasonable discovery to ascertain whether there is a hidden purpose that motivated the issuance of the grant. Discovery, however, should not be any broader or any more intrusive than it needs to be. For the purpose of ascertaining the purpose of the grants, discovery should be limited to the testimony of the rule 30 (b) (6) witness and writings reflecting the oral and written communications regarding the committee's decision-making process in recommending the grants; there is no need in this case to probe the private intentions of town meeting members. We leave it to the judge in her discretion to determine more precisely the appropriate scope of discovery.
b. Substantial aid. The second factor to be considered is whether the effect of the grants is to substantially aid a *707church. Our precedents make clear that a grant of public funds does not violate the anti-aid amendment if the assistance it provides to a private institution is merely "minimal," Essex, 387 Mass. at 332, 439 N.E.2d 770, or "remote," Bloom, 376 Mass. at 47, 379 N.E.2d 578. The aid must provide "substantial **89assistance" to the church to risk violation of the anti-aid amendment. Springfield, 382 Mass. at 680, 417 N.E.2d 408. In evaluating this factor, we look to both the amount of aid provided and "the degree to which the aid assists [the church] in carrying out [its] essential function." Opinion of the Justices, 401 Mass. at 1208, 514 N.E.2d 353.
In particular, we have focused on whether the aid that is provided contains certain "limiting features" designed to restrict its effect. Id. at 1207, 514 N.E.2d 353. In Springfield, we approved the funding of the special education programs with the important limitation that there would be no reimbursement for children whose parents had unilaterally enrolled them in private school; public funding was strictly limited to expenses that the private schools would not otherwise have incurred. See Springfield, 382 Mass. at 677, 417 N.E.2d 408. This limiting feature worked to cabin the effect of the public funding, guaranteeing that it would not "aid the private school [s] in carrying out [their] essential function." Id. at 681, 417 N.E.2d 408.
We see no such guarantee here. As an initial matter, we note that the proposed grants are "neither minimal nor insignificant" in amount. Opinion of the Justices, 401 Mass. at 1208, 514 N.E.2d 353. The total cost of the comprehensive assessment contemplated under the Master Plan will be $55,000, to which the Master Plan grant will contribute $49,500, while the total cost of restoring the stained glass windows will be $56,930, to which the stained glass grant will contribute $51,237.
More worrisome is the extent to which these grants will assist the church in its "essential enterprise" as an active house of worship. Bloom, 376 Mass. at 47, 379 N.E.2d 578. The church was candid in its grant applications, explaining that-faced with declining membership and contributions-it would need the town's "help" in order to preserve its buildings while also "offering the congregation what draws them to their church." This is not a case like in Springfield, where it was possible to limit the public funding to a narrow, specific purpose. The reimbursement there was for expenses that the schools would not otherwise have incurred; it did nothing to "lessen[ ] the financial burden" of the schools or those who chose to attend those schools. Springfield, 382 Mass. at 683, 417 N.E.2d 408. Here, in contrast, the grants would help defray planning and restoration costs that the church would otherwise have to shoulder on its own, allowing the money saved to be used to support its core religious activities. As the church indicated in its grant applications, budgetary constraints have led it to make difficult choices between "capital improvement projects" on the one hand **90and "programs and personnel" on the other. These grants would allow the church to have both, in effect "underwrit[ing]" its essential function as an active house of worship. Opinion of the Justices, 401 Mass. at 1209, 514 N.E.2d 353.
On this record, we therefore conclude that the effect of these grants is to substantially aid the church.
c. Risks. The third and last factor that must be considered is whether the grants avoid the risks that prompted the passage of the anti-aid amendment. In evaluating the third factor, the judge erred in focusing on whether there was "credible *708evidence that the grants under the [act] are economically or politically abusive or unfair," and, finding no such evidence, concluding that there was "no political or economic abuse which the anti-aid amendment was enacted to prevent." Instead, the judge should have focused on whether the grants to the church avoid the risks of the political and economic abuses that "prompted the passage" of the anti-aid amendment. Springfield, 382 Mass. at 675, 417 N.E.2d 408.
We recognize that our articulation of this third factor in prior cases has provided less than clear guidance. The third factor, as first set forth in Springfield, focused on "whether the [grant] avoids the political and economic abuses which prompted the passage of [the anti-aid amendment]." Id. But in Springfield, we did not provide the historical background that identified these "political and economic abuses," and therefore failed to recognize, as we do here, that the amendment was proposed in 1853 not to abolish an existing practice of funding religious institutions-no one at the Convention alleged the existence of such a practice-but instead as a preemptive measure to avoid the risks associated with the public financial support of religious institutions. These risks, as we noted in Bloom, 376 Mass. at 39, 379 N.E.2d 578, also prompted the revision of the anti-aid amendment in 1917, and are worth repeating here: first, the risk that "liberty of conscience" would be infringed "whenever a citizen was taxed to support the religious institutions of others"; second, the risk that public funding would result in improper government entanglement with religion, undermining the "independence and dignity" of churches; and third, the risk that the public support of religious institutions would threaten "civic harmony," making the divisive "question of religion" a political question. Id.
In Helmes, 406 Mass. at 878, 550 N.E.2d 872, our most recent case applying the three-factor test, we redefined the third factor in light of the circumstances of that case to consider "whether there is any use **91of public money that aids a charitable undertaking in a way that is abusive or unfair, economically or politically." Because nothing in the record indicated any such abuse or unfairness, we concluded that the appropriation was constitutional; there was no evidence that any private person would benefit from it, that the funds would be distributed to a noncharitable use, or that its charitable objective-preserving a World War II battleship and educating the public-was not generally accepted. Id. at 877-878, 550 N.E.2d 872. We did not consider in Helmes whether the appropriation of funds presented any of the risks that the framers of the anti-aid amendment sought to avoid, perhaps because it was so clear that these risks were not presented where the challenged funding was for the repair of a memorial battleship.
Here, where the grant of public funds is for the renovation of an active house of worship, it is imperative, in considering the third factor, to focus on whether these specific grants avoid the risks of the political and economic abuses that "prompted the passage" of the anti-aid amendment, which we identified in Bloom and have described in this opinion. On the record before us, we conclude that these risks are significant.
First, these grants risk infringing on taxpayers' liberty of conscience-a risk that was specifically contemplated by the framers of the anti-aid amendment. As one delegate to the Convention of 1917 stated, "Religious liberty [requires] that ... the State cannot compel a man to pay his good money in taxation for the support of a religion, or of the schools and institutions *709of a religion, in which he does not believe." Debates of 1917-1918, supra at 77. The self-described mission of the church here is "to preach and teach the good news of the salvation that was secured ... through the life, death, and resurrection of Jesus." The proposed grants would be used to renovate the main church building, where the church conducts its worship services, and its stained glass windows, which feature explicit religious imagery and language. For town residents who do not subscribe to the church's beliefs, the grants present a risk that their liberty of conscience will be infringed, especially where their tax dollars are spent to preserve the church's worship space and its stained glass windows.
Second, these grants also present a risk of government entanglement with religion. See Bloom, 376 Mass. at 39, 47, 379 N.E.2d 578. To ensure that the grants are used for historic preservation, the town has imposed on the church the condition that it execute a historic **92preservation restriction, which-if the restrictions accompanying the town's prior grants under the act are any indication-would significantly limit the church's ability to make future alterations to its buildings, including its worship space and its stained glass windows, without the town's approval.23 We have held in other contexts that where the State exercises control over the design features of a church, it infringes on the free exercise of religion guaranteed under the Massachusetts Constitution. In The Society of Jesus of New England v. Boston Landmarks Comm'n, 409 Mass. 38, 42, 564 N.E.2d 571 (1990) ( Society of Jesus ), we concluded that the designation of a church interior as a landmark, thereby making all renovations subject to government approval, infringed on "the right freely to design interior spaces for religious worship," in violation of art. 2 of the Massachusetts Declaration of Rights. The historic preservation restriction contemplated here presents a comparable risk of "intrusion ... , reaching into the church's actual worship space." Id.
The town contends that these grants would result in no such intrusion, and are distinguishable from the landmark designation in Society of Jesus, because they relate only to the exterior of the church's buildings. See, e.g., G. L. c. 40C, § 7 ("The [historic district] commission shall not consider interior arrangements or architectural features not subject to public view"). In Society of Jesus, 409 Mass. at 39 n.2, 564 N.E.2d 571, we expressly did not decide whether a landmark designation of a church exterior would also infringe on the free exercise of religion. We need not decide that issue here because, even if we were to recognize the distinction between the interior and exterior of a church and conclude that restrictions on the renovation of a church exterior would not burden the free exercise of religion, such restrictions would still pose a risk of government entanglement in religious matters.
In Society of Jesus, we reasoned that "[t]he configuration of the church interior is so freighted with religious meaning that it must be considered part and parcel of ... religious worship." Society of Jesus, 409 Mass. at 42, 564 N.E.2d 571. Since then we have recognized that the exterior features of a religious structure can also be expressive of **93religious beliefs. In *710Martin v. The Corporation of the Presiding Bishop of the Church of Jesus Christ of Latter-Day Saints, 434 Mass. 141, 142, 747 N.E.2d 131 (2001), we held that a church steeple should be exempted from local height restrictions as a "religious" use of land, noting that "churches have long built steeples to 'express elevation toward the infinite' " (citation omitted). Id. at 152, 747 N.E.2d 131. See P. Tillich, On Art and Architecture 212 (1989) ("the one great symbol of the church building is the building itself"). We warned, "It is not for judges to determine whether the inclusion of a particular architectural feature is 'necessary' for a particular religion," Martin, supra at 150, 747 N.E.2d 131, or "to determine what is or is not a matter of religious doctrine." Id. at 152, 747 N.E.2d 131. The Master Plan grant at issue here contemplates a comprehensive assessment of the entire church building, which would include elements both exterior and interior; it is not for judges or, for that matter, a community preservation committee to determine whether this assessment will affect elements that touch on matters of religious doctrine.
The stained glass window is illustrative of the fragility of the interior-exterior distinction, and of the extent to which historic preservation of the building is interwoven with religious doctrine. Although it is an "exterior" feature, in that it is open to public view, see G. L. c. 40C, § 5, its inclusion in a church building is as much a religious choice as an aesthetic one-especially where, as here, the windows have an expressly religious message. See V.C. Raguin, Stained Glass, From its Origins to the Present, 10-13 (2003).
Third, the challenged grants also risk threatening "civic harmony," by making the "question of religion" a political one. Bloom, 376 Mass. at 39, 379 N.E.2d 578. As centuries of experience have shown, government support of churches has always and inevitably been a politically divisive issue in Massachusetts. Although the act provides for a rigorous process for the allocation of funds, the decision to award a grant lies with the committee and, ultimately, with the town meeting members. Those who first proposed the anti-aid amendment in 1853 were wary of throwing "a firebrand into ... town meetings." Debates of 1853, supra at 624. Grants for the renovation of churches-using funds that could potentially have been dedicated to open space, soccer fields, low-income housing, or other historic preservation projects, including projects for the renovation of houses of worship of other religious denominations-pose an inevitable risk of making "the irritating **94question of religion" a politically divisive one in a community, the more so where those grants are for the renovation of a worship space or of a stained glass window with explicit religious imagery. Bloom, supra at 39, 379 N.E.2d 578.
We do not suggest that fair consideration of the risks that prompted the passage of the anti-aid amendment means that every historic preservation grant for a church building will be unconstitutional. We only caution that any such grant to an active church warrants careful scrutiny under the three-factor Springfield test. The third factor is by no means a dispositive factor, only an important one. Indeed, we can imagine various circumstances where such grants would survive careful scrutiny, including, for instance, where historical events of great significance occurred in the church, or where the grants are limited to preserving church property with a primarily secular purpose. Cf. Shrine of Our Lady of La Salette Inc. v. Board of Assessors of Attleboro, 476 Mass. 690, 700-702, 71 N.E.3d 509 (2017) (shrine property leased for battered women's shelter and used as wildlife sanctuary not subject to religious worship exemption, because "dominant *711purpose" not connected to religious worship and instruction). The use of public funds for such preservation efforts poses little risk of political division.24 **95In this case, having weighed and balanced the three factors, we conclude that the plaintiffs are likely to succeed on the merits of their claim with respect to the stained glass grant. Although the record before us does not allow us to ascertain whether there is a motivating purpose behind this grant other than historic preservation, its effect is to substantially aid the church in its essential function and, given the explicit religious imagery of the stained glass, it fails to avoid the very risks that the framers of the anti-aid amendment hoped to avoid. Thus, even if further discovery were to reveal that the sole motivating purpose of this grant was in fact to preserve historic resources, and not to aid this particular church, the other factors in our analysis-especially the third factor, to which we accord special weight-still compel the conclusion that the stained glass grant runs afoul of the anti-aid amendment. Because the plaintiffs are likely to succeed on the merits of their claim, and a preliminary injunction would "promote[ ] the public interest" reflected in the anti-aid amendment, LeClair, 430 Mass. at 332, 719 N.E.2d 464, the plaintiffs are entitled to a preliminary injunction barring the disbursement of the stained glass grant.
With respect to the Master Plan grant, we conclude that further discovery is needed before a determination should be made as to whether the plaintiffs are likely to succeed on the merits of their claim. This is in part because, unlike the stained glass grant, the Master Plan grant is far broader in its scope, including not only plans for the renovation of worship space but also plans for the renovation of the Fletcher and Hosmer Houses, which are both private residences. Accordingly, analysis of the grant under the third factor must be more fact-intensive; restoration of the main church building will implicate risks different from those arising from the *712restoration of the adjoining residences. And where the analysis of the third factor is more complex, and the potential judicial options more diverse,25 the discovery that might shed light on whether there was a hidden purpose apart from historic preservation becomes more important to the over-all decision.
We therefore remand the issue to the Superior Court for a determination whether the Master Plan grant, in full or in part, **96should survive the careful scrutiny required under the third factor. Such a determination should not be made until the plaintiffs have had reasonable discovery regarding the purpose of the committee in awarding this grant. We reiterate that the scope of such discovery should be limited at this time to the testimony of the rule 30 (b) (6) witness and writings reflecting the oral and written communications regarding the committee's decision-making process in recommending the grants and that there is no need to probe the private intentions of town meeting members. We leave it to the judge to determine more precisely its appropriate scope.
Conclusion. The orders denying the plaintiffs' motion for a preliminary injunction and granting the town's motion for a protective order to stay discovery are vacated. The case is remanded to the Superior Court for entry of an order allowing the plaintiffs' motion for a preliminary injunction barring disbursement of the stained glass grant and, as to the Master Plan grant, for further proceedings consistent with this opinion.
So ordered.
KAFKER, J. (concurring, with whom Gaziano, J., joins).
I write separately to emphasize that our analysis of the anti-aid amendment of the Massachusetts Constitution is tightly constrained by the United States Supreme Court's interpretation of the religion clauses of the First Amendment to the United States Constitution. The grants at issue here are provided pursuant to a generally available public benefit program designed to promote community conservation including the protection of the Commonwealth's historic buildings. The United States Supreme Court has warned that only a very narrow category of exclusions are allowed by the free exercise clause from such generally available public benefit programs. Because I believe the preliminary injunction against the stained glass grant is consistent with this very narrow permitted exclusion, and the Master Plan grant requires further analysis to decide both the anti-aid and First Amendment questions, I concur in the judgment of the court.
1. The First Amendment background to this case. Today's decision takes us into one of the most confusing and contested areas of State and Federal constitutional law. The United States Supreme Court has emphasized that there is a "tension" between the religion clauses of the United States Constitution-that is, **97what is prohibited by the establishment clause and what is required by the free exercise clause of the First Amendment. See Locke v. Davey, 540 U.S. 712, 718, 124 S.Ct. 1307, 158 L.Ed.2d 1 (2004). The Court has also stated that there is "play in the joints" between the dictates of the two religion provisions in the United States Constitution-allowing limited State action therein-without defining precisely how much play. See Trinity Lutheran Church of Columbia, Inc. v. Comer, --- U.S. ----, 137 S.Ct. 2012, 2019, 198 L.Ed.2d 551 (2017) *713( Trinity Lutheran ). The Supreme Court's jurisprudence also has been continually evolving, particularly in its definition of the neutrality the two first amendment provisions requires in regard to religion.1
All of this is further complicated by State constitutional anti-aid provisions providing greater protections against the establishment of religion than the establishment clause of the First Amendment. These State constitutional anti-aid provisions present additional legal constraints, and State grants are permissible only if they do not run afoul of the free exercise clause of the First Amendment.
There is no clear path yet through this difficult intersection of the religion clauses of the State and Federal Constitutions. Most instructive, for our purposes, however, are the Supreme Court's more recent pronouncements in Trinity Lutheran and Locke. These two cases analyzed grants arising from generally available public benefit programs, like the one before us. See Trinity Lutheran, supra at 2017 ; Locke, supra at 715, 124 S.Ct. 1307. Both cases involved exclusions required by anti-aid provisions in State Constitutions. See Trinity Lutheran, supra at 2017 ( Missouri Constitution, art. 1, § 7 ); Locke, supra at 722, 124 S.Ct. 1307 ( Washington Constitution, art. 1, § 11 ).
In Trinity Lutheran, 137 S.Ct. at 2025, the Supreme Court held **98that the exclusion of a church school and day care facility from a generally available public benefit program funding rubber playground surfaces "solely" on account of a church's religious identity violated the free exercise clause. The Court held that it had "repeatedly confirmed" that it will not approve such exclusions, giving as an example its 1947 decision upholding against Federal establishment clause challenges a New Jersey law allowing a local school district to pay for public, private, and parochial school transportation costs. Id. at 2019-2020, citing Everson v. Board of Educ. of Ewing, 330 U.S. 1, 67 S.Ct. 504, 91 (L.Ed. 711 1947).
In Locke, however, the Supreme Court held that a State anti-aid amendment exclusion of scholarships to pursue degrees in devotional theology from an otherwise inclusive student aid program did not violate the free exercise clause of the First Amendment. Locke, 540 U.S. at 725, 124 S.Ct. 1307. In so holding, the Court stressed that it could "think of few areas in which a State's antiestablishment interests come more into play" than using "taxpayer funds to support church leaders." Id. at 722, 124 S.Ct. 1307. "The claimant in Locke sought funding for an 'essentially religious endeavor ... akin to a religious calling.' " Trinity Lutheran, 137 S.Ct. at 2023, quoting Locke, supra at 721-722, 124 S.Ct. 1307. To contrast, the Court in Trinity Lutheran stated, "nothing of the sort can be said about a program to use recycled tires to resurface playgrounds." Trinity Lutheran, supra. In his concurrence in *714Trinity Lutheran, Justice Breyer also emphasized that he would "find relevant, and would emphasize, the particular nature of the 'public benefit' ... at issue." Id. at 2026 (Breyer, J., concurring).
Together, Trinity Lutheran and Locke define a very narrow category of exclusions from generally available public benefit programs that can be required by State anti-aid amendments without violating the free exercise clause of the First Amendment. To be excluded from a generally available public benefit program, the funding must be sought for an "essentially religious endeavor" raising important state constitutional antiestablishment concerns. Trinity Lutheran, 137 S.Ct. at 2023, quoting Locke, 540 U.S. at 721-722, 124 S.Ct. 1307. With these overarching First Amendment principles in mind, I turn to the grants at issue, and art. 18 of the Amendments to the Massachusetts Constitution, as amended by arts. 46 and 103 of the Amendments, the anti-aid amendment.
2. The Community Preservation Act grant and the anti-aid amendment. As explained by the court, the town of Acton (town) is one of 172 municipalities in Massachusetts that have adopted **99the Community Preservation Act (act), which establishes processes and procedures for funding projects related to open space, historic resources, and community housing. See ante at 72, 92 N.E.3d at 694-695. Here, the church's "Evangelical Church Stained Glass Window Preservation" application initially requested $41,000 from the town's Community Preservation Committee (committee) to repair the church's stained glass windows. Eventually $51,237 was awarded for the windows. The proposed repairs included a three-foot, six-inch by ten-foot, six-inch "Christ window" depicting Jesus with a woman kneeling and praying, altar windows, and a window containing a cross and the hymnal phrase "Rock of Ages Cleft for Me."2 The church was requesting that the town pay for ninety per cent of the costs. The stained glass windows were "installed in memorial to honor prominent members of the church" in 1898.
The church also sought $49,500 to hire an architect to do a structural review and prepare a master plan for historic preservation of the church, and two neighboring buildings owned by the church, the John Fletcher House and the Abner Hosmer House. The church was again requesting that the town pay ninety per cent of the costs. The main church dates back to 1846 with a renovation in 1898. The houses were built circa 1855 and 1846. The grant was sought to "hire an architectural consultant to thoroughly investigate each of the [three] historic buildings to identify all the needs of each building in order to protect and preserve these historic assets for future generations." For the church itself, this would include "a thorough assessment of the [c]hurch building envelope, including windows, doors, siding, roof, chimney, bell tower, skylights, and fire escapes, with a focus on protecting the building from the elements." Similarly, "the rental houses will be evaluated for the building envelope, mechanical, electrical and plumbing systems, and safety systems. This work will focus on building structural integrity." The grant was requested because "each [of the *715buildings] shows the signs of 170+ years of wear."
In its application for both grants, the church explained that "mainstream churches have not been growing for years, and the **100financial strain is significant ... we have had to cut programs and personnel. The cuts can further exacerbate the financial problem[s] by not offering the congregation what draws them to their church."
Pursuant to the requirements of the act, the committee held a public hearing and voted unanimously to recommend the grants. The town meeting approved both grants. The annual town meeting warrant explained that the church and the other two buildings were located in the Acton Centre Historic District. The warrant explained that the "work will protect the stained glass windows, an integral part of the church's historical significance." The warrant also explained that the master plan would evaluate and identify critical needs and set restoration and rehabilitation priorities to preserve the three historic buildings. It also stated that the "preservation project must comply with the Standards for Rehabilitation stated in the United States Secretary of the Interior's Standards for the Treatment of Historic Properties codified in 36 C.F.R. Part 68." Historic preservation restrictions were imposed on the buildings with the restriction being "perpetual to the extent permitted by law." The plaintiffs, who are town taxpayers, challenged the grants, claiming they violate the anti-aid amendment.
3. Application of the anti-aid amendment and the First Amendment to the stained glass grant. I agree with the court that the three factor anti-aid amendment analysis set forth in Commonwealth v. School Comm. of Springfield, 382 Mass. 665, 675, 417 N.E.2d 408 (1981) ( Springfield ), applies, including where the grant is being given to a church as well as a nonreligious private charity. I also agree that a categorical ban would violate the First Amendment right to the free exercise of religion.
In analyzing the first factor, I conclude that we must consider the purpose of both the statute and the grant. This is necessitated, in part, by the Supreme Court's First Amendment jurisprudence and its focus on whether the grant is authorized pursuant to a generally available public benefit program. Here, the purpose of the statute itself is unquestionably to provide generally available public benefits for the purpose of conservation, including historic preservation. There is no suggestion or argument that an "examination of the statutory scheme ... [will reveal] any 'technique of circumvention' " designed to avoid the requirements of the anti-aid amendment. Springfield, 382 Mass. at 677, 417 N.E.2d 408, quoting Bloom v. School Comm. of Springfield, 376 Mass 35, 47, 379 N.E.2d 578 (1978). See **101Bloom, supra at 44, 379 N.E.2d 578 ("[W]e note, first, that the Supreme Court has been regularly unreceptive to schemes of circumvention which resemble that attempted by the present legislation"). Indeed the statute is straightforward and serves important conservation purposes as eloquently explained by the dissent. See post at 111-112, 92 N.E.3d at 723-724.
The court, however, draws a distinction between the purposes of the statute and those of the grants, and emphasizes that we must probe further to discern the primary or motivating purposes of the grantors as well as any hidden purposes, and this additional inquiry requires a remand for the Master Plan grant. See ante at 95-96, 92 N.E.3d at 711-712. At least for a determination whether a preliminary injunction should issue regarding the stained glass grant, I conclude that we have a sufficient record that conservation is the primary purpose of the *716grants. I do not detect any indicia of a scheme or technique of circumvention. The purpose, as reflected in the town warrant, appear to be described straightforwardly and factually.
In my opinion, the most complicated aspect of the purpose inquiry is not discerning the subjective intentions of the grantors but the difficulty of separating conservation from religious purposes when the grant is being given to preserve a religious component of a church building. Even if the purpose of the grantors is conservation, and not the promotion of religion, it is obvious to anyone voting on the grants that both purposes would be served. I think that is particularly true for the stained glass grant where the windows convey an express sectarian religious message.3 Ultimately, however, the purpose inquiry is just one **102factor in a multifactor test and it is meant to be instructive, not dispositive. Springfield, 382 Mass. at 675, 417 N.E.2d 408. I find the other two factors, particularly the third, conclusive of the anti-aid amendment analysis and critical to the First Amendment interpretation as well.
The second prong of the anti-aid test analyzes whether the grants substantially assist religion. The stained glass grant is "neither minimal nor insignificant" to the church. See Opinion of the Justices, 401 Mass. 1201, 1208, 514 N.E.2d 353 (1987). Approximately $50,000 is being provided and the town is funding ninety per cent of the total cost. Without the assistance of the committee's grants, the church indicated that the financial strain and required cuts could "exacerbate the financial problem[s] by not offering the congregation what draws them to their church."4
*717Most important in my view is the third prong. Awarding public monies paid by taxes directly to a church to repair stained glass windows with an express religious message raises core concerns about separation of church and State that prompted the passage of the anti-aid amendment. I agree with the court that those concerns include (1) infringement on liberty of conscience caused by taxing citizens to support the religious beliefs and institutions of **103others; (2) improper government entanglement with religion, thereby diminishing the independence and integrity of both church and State; and (3) unnecessary divisiveness in the polity caused by making the funding of religious institutions a political question. See ante at 91-94, 92 N.E.3d at 708-711.
All three of these risks are present here. Tax dollars are paying for the stained glass windows that have an express sectarian religious message. A historic preservation restriction of perpetual duration is being imposed on the windows and perhaps other parts of the church, thereby entwining an active church building with s tate government. See The Society of Jesus of New England v. Boston Landmarks Comm'n, 409 Mass. 38, 42, 564 N.E.2d 571 (1990) (designation of church interior as landmark infringed on "right freely to design interior spaces for religious worship"). See also Martin v. The Corporation of thePresiding Bishop of the Church of Jesus Christ of Latter-Day Saints, 434 Mass. 141, 153, 747 N.E.2d 131 (2001) ("no municipal concern was served by controlling the steeple height of churches"); Saperstein, Public Accountability and Faith-Based Organizations: A Problem Best Avoided, 116 Harv. L. Rev. 1353, 1365 (2003) ("With government money come government rules, regulations, audits, monitoring, interference, and control-all of which inherently threaten religious autonomy"). Town meeting members were being asked to vote on a grant to maintain religious aspects of the church of their neighbors and now they are suing each other. Should another house of worship in the town be denied a grant after this one has been awarded, it will likely bring about further controversy and division. No more discovery is required to know that this grant goes to core concerns of the anti-aid amendment.5 In sum, the balancing of the three factors shows that the plaintiffs have a substantial likelihood of success in establishing that the stained glass grant violates the anti-aid amendment.
As the church and the free exercise rights of its members are also implicated, they must be considered as well. As explained above, to be excluded from a generally available public benefit program, the funding must be sought for an "essentially religious endeavor" raising important State constitutional antiestablishment concerns. See Locke, 540 U.S. at 721, 124 S.Ct. 1307. I conclude that paying **104for stained glass windows with an express sectarian religious message and mission fits *718within the very narrow exception allowed by Locke.
The benefits are vastly different from the nonreligious rubberized playground services or school transportation costs, or the police and fire or other obviously nonreligious types of assistance that have been found not to raise establishment clause or anti-aid concerns. See Trinity Lutheran, 137 S.Ct. at 2026-2027 (Breyer, J., concurring). See also Everson, 330 U.S. at 17-18, 67 S.Ct. 504 (describing services "so separate and so indisputably marked off from the religious function"). Although "nothing [religious] ... can be said about a program to use recycled tires to resurface playgrounds," the opposite is true for stained glass windows. See Trinity Lutheran, supra at 2023. They are an important part of the church's religious message and mission. V.C. Raguin, Stained Glass, From its Origins to the Present, 13 (2003) ("stained glass became ... an intimation of God's very nature, and important as a contemplative aid"); Lupu & Tuttle, Historic Preservation Grants to House of Worship: A Case Study in The Survival of Separationism, 43 B.C. L. Rev. 1139, 1175 (2002) ("[Stained glass] windows often present religious themes ... and help to shape the worship experience through the play of light and imagery"). See Mitchell v. Helms, 530 U.S. 793, 820, 120 S.Ct. 2530, 147 L.Ed.2d 660 (2000) (opinion of Thomas, J.) (aid cannot be "impermissibly religious in nature"). Additionally, as explained above, the stained glass grant here raises core State constitutional anti-aid concerns. Like excluding State scholarships to pay for a divinity degree in Locke, there are "few areas in which a State's antiestablishment interests come more into play" than paying for stained glass windows with sectarian religious symbolism. Locke, 540 U.S. at 722, 124 S.Ct. 1307.
For the religion clauses in the State and Federal Constitutions, there is "no simple and clear measure which by precise application can readily and invariably demark the permissible from the impermissible." School Dist. of Abington Township, Pa. v. Schempp, 374 U.S. 203, 306, 83 S.Ct. 1560, 10 L.Ed.2d 844 (1963) (Goldberg, J., concurring). See Van Orden v. Perry, 545 U.S. 677, 699, 125 S.Ct. 2854, 162 L.Ed.2d 607 (2005) (Breyer, J., concurring) ("the Court has found no single mechanical formula that can accurately draw the constitutional line in every case"). Although line drawing in this intensely contested area of constitutional law is difficult, I believe that the use of taxpayer dollars to pay for stained glass windows with a religious message crosses that line.
**105I therefore conclude that on this record the plaintiffs have demonstrated the necessary likelihood of success that the stained glass grant violates the State's anti-aid amendment without running afoul of the free exercise clause.
4. Remand on the Master Plan grant. I also agree with the court that a remand is required on the Master Plan grant, although I place less emphasis than the court on a search for "hidden" purposes. I conclude that a fuller factual record is required on the inner workings of the grant itself before it can be determined whether the Master Plan grant violates the anti-aid amendment, and if so, whether exclusion of such a grant from a generally available public benefit program would violate the free exercise clause of the First Amendment.
It is important to emphasize up front just how narrow the exclusion is for generally available public benefit programs. See Locke, 540 U.S. at 725, 124 S.Ct. 1307. The exclusion involves essentially religious endeavors, such as paying for ministry training or stained glass windows with sectarian *719symbols or messages. The Master Plan grant is to pay an architect to perform a structural review of three 170 year old buildings of historic importance to the town. Only one of those buildings is a church. The focus of the architect's work appears to be on preserving the structural integrity of the old buildings, not repairing or maintaining particular parts of the church that convey an express religious message.6 It is unclear to me how much of this work goes beyond the "building envelope." These buildings are also a part of the historic district of the town and serve important nonreligious as well as religious purposes in the town and the Commonwealth, as the dissent explains. See post at 111-112, 92 N.E.3d at 723-724. Additionally it is not clear from the record what historic preservation restriction will result from this grant. Will the grant to pay for an architect to provide for a structural review of the three buildings give the town a restriction regarding construction on all of these buildings? Or would such a restriction only apply if a grant is provided for subsequent work **106on the buildings? A fuller factual record is necessary on this point as well as others.
5. Conclusion. In sum, I conclude that the stained glass grant not only violates the anti-aid amendment but also fits within the very narrow exclusion from a generally available public benefit program authorized by the Supreme Court pursuant to the First Amendment. I further conclude that on remand the legal status of the Master Plan grant under both the anti-aid amendment and the free exercise clause of the First Amendment must be determined.