Plaintiffs brought this action under the Uniform Declaratory Judgments Act, ORS 28.010 - 28.160, challenging the constitutionality of Ballot Measure 3 (2000) (Measure 3 or Oregon Property Protection Act).1 Measure 3 added section 10 to Article XV of the Oregon Constitution. Section 10 consists of 12 subsections. Plaintiffs argued that Measure 3 contains two or more constitutional amendments that should have been voted on separately under Article XVII, section 1, of the Oregon Constitution.2 Plaintiffs also argued that Measure 3 addresses more than one subject, in violation of Article IV, section 1(2)(d), of the Oregon Constitution.3

The circuit court entered judgment for defendants and upheld the constitutionality of Measure 3 against both challenges. For the reasons that follow, we reverse the judgment of the circuit court.

Plaintiff Lincoln County is a political subdivision of thé State of Oregon, and plaintiff Lincoln Interagency Narcotics Team is a law enforcement agency created by intergovernmental agreement under ORS chapter 190. Both plaintiffs are subject to rights and responsibilities under Measure 3. Plaintiff intervenors are three individuals and three private organizations that are involved in various ways with animal protection issues.4 Defendants are the former Governor and the Secretary of State of Oregon. Defendant *530intervenors are the chief petitioners for Measure 3. Because this action presented only legal issues concerning the constitutionality of Measure 3, the parties filed cross-motions for summary judgment. After the trial court entered summary judgment for defendants, plaintiffs appealed to this court.

We begin by describing the history and substance of Measure 3. The voters adopted the measure on November 7, 2000. Measure 3 provides:

“Article XV of the Constitution of the State of Oregon is amended by a vote of the People to include the following new section:

“Section 10. The Oregon Property Protection Act of2000. (1) This section may be known and shall be cited as the ‘Oregon Property Protection Act of 2000.’

“(2) Statement of principles. The People, in the exercise of the power reserved to them under the Constitution of the State of Oregon, declare that:

“(a) A basic tenet of a democratic society is that a person is presumed innocent and should not be punished until proven guilty;

“(b) The property of a person should not be forfeited in a forfeiture proceeding by government unless and until that person is convicted of a crime involving the property;

“(c) The value of property forfeited should be proportional to the specific conduct for which the owner of the property has been convicted; and

“(d) Proceeds from forfeited property should be used for treatment of drug abuse unless otherwise specified by law for another purpose.

“(3) Forfeitures prohibited without conviction. No judgment of forfeiture of property in a civil forfeiture proceeding by the State or any of its political subdivisions shall be allowed or entered until and unless the owner of the property is convicted of a crime in Oregon or another jurisdiction and the property is found by clear and convincing evidence to have been instrumental in committing or facilitating the crime or to be proceeds of that crime. The value of the property forfeited under the provisions of this subsection shall not be excessive and shall be substantially proportional to the specific conduct for which the owner of *531the property has been convicted. For purposes of this section, ‘property’ means any interest in anything of value, including the whole of any lot or tract of land and tangible and intangible personal property, including currency, instruments or securities or any other kind of privilege, interest, claim or right whether due or to become due. Nothing in this section shall prohibit a person from voluntarily giving a judgment of forfeiture.

“(4) Protection of innocent property owners. In a civil forfeiture proceeding if a financial institution claiming an interest in the property demonstrates that it holds an interest, its interest shall not be subject to forfeiture.

“In a civil forfeiture proceeding if a person claiming an interest in the property, other than a financial institution or a defendant who has been charged with or convicted of a crime involving that property, demonstrates that the person has an interest in the property, that person’s interest shall not be subject to forfeiture unless:

“(a) The forfeiting agency proves by clear and convincing evidence that the person took the property or the interest with the intent to defeat the forfeiture; or

“(b) A conviction under subsection (3) is later obtained against the person.

“(5) Exception for unclaimed property and contraband. Notwithstanding the provisions of subsection (3) of this section, if, following notice to all persons known to have an interest or who may have an interest, no person claims an interest in the seized property or if the property is contraband, a judgment of forfeiture may be allowed and entered without a criminal conviction. For purposes of this subsection, ‘contraband’ means personal property, articles or things, including but not limited to controlled substances or drug paraphernalia, that a person is prohibited by Oregon statute or local ordinance from producing, obtaining or possessing.

“(6) Law enforcement seizures unaffected. Nothing in this section shall be construed to affect the temporary seizure of property for evidentiary, forfeiture, or protective purposes, or to alter the power of the Governor to remit fines or forfeitures under Article V, Section 14, of this Constitution.

*532“(7) Disposition of property and proceeds to drug treatment. Any sale of forfeited property shall be conducted in a commercially reasonable manner. Property or proceeds forfeited under subsections (3), (5), or (8) of this section shall not be used for law enforcement purposes but shall be distributed or applied in the following order:

“(a) To the satisfaction of any foreclosed liens, security interests and contracts in the order of their priority;

“(b) To the State or any of its political subdivisions for actual and reasonable expenses related to the costs of the forfeiture proceeding, including attorney fees, storage, maintenance, management, and disposition of the property incurred in connection with the sale of any forfeited property in an amount not to exceed twenty-five percent of the total proceeds in any single forfeiture;

“(c) To the State or any of its political subdivisions to be used exclusively for drug treatment, unless another disposition is specially provided by law.

“(8) State and federal sharing. The State of Oregon or any of its political subdivisions shall take all necessary steps to obtain shared property or proceeds from the United States Department of Justice resulting from a forfeiture. Any property or proceeds received from the United States Department of Justice by the State of Oregon or any of its political subdivisions shall be applied as provided in subsection (7) of this section.

“(9) Restrictions on State transfers. Neither the State of Oregon, its political subdivisions, nor any forfeiting agency shall transfer forfeiture proceedings to the federal government unless a state court has affirmatively found that:

“(a) The activity giving rise to the forfeiture is interstate in nature and sufficiently complex to justify the transfer;

“(b) The seized property may only be forfeited under federal law; or

“(c) Pursuing forfeiture under state law would unduly burden the state forfeiting agencies.

“(10) Penalty for violations. Any person acting under color of law, official title or position who takes any *533action intending to conceal, transfer, withhold, retain, divert or otherwise prevent any proceeds, conveyances, real property, or any things of value forfeited under the law of this State or the United States from being applied, deposited or used in accordance with subsections (7), (8) or (9) of this section shall be subject to a civil penalty in an amount treble the value of the forfeited property concealed, transferred, withheld, retained or diverted. Nothing in this subsection shall be construed to impair judicial immunity if otherwise applicable.

“(11) Reporting requirement. All forfeiting agencies shall report the nature and disposition of all property and proceeds seized for forfeiture or forfeited to a State asset forfeiture oversight committee that is independent of any forfeiting agency. The asset forfeiture oversight committee shall generate and make available to the public an annual report of the information collected. The asset forfeiture oversight committee shall also make recommendations to ensure that asset forfeiture proceedings are handled in a manner that is fair to innocent property owners and interest holders.

“(12) Severability. If any part of this section or its application to any person or circumstance is held to be invalid for any reason, then the remaining parts or applications to any persons or circumstances shall not be affected but shall remain in full force and effect.”

(Boldface in original.)

In their “separate vote” challenge under Article XVII, section 1, plaintiffs argue that Measure 3 presented to the voters, in a single measure, an amendment that included at least eight separate changes to the Oregon Constitution.5 Among those changes, plaintiffs contend that Measure 3 made, by addition, the following seven changes to Article XV, of the Oregon Constitution:

(1) Subsection (3), generally requiring that the property owner be convicted of a crime as a prerequisite to the civil forfeiture of property;

*534(2) Subsection (3), establishing the burden of proof in civil forfeiture actions as “clear and convincing evidence”;

(3) Subsection (4), prohibiting civil forfeiture of the property of a person who has not been convicted of a crime unless “[t]he forfeiting agency proves by clear and convincing evidence that the person took the property or the interest with the intent to defeat the forfeiture”;

(4) Subsection (7), prohibiting the use of Oregon civil forfeiture proceeds for law enforcement purposes and, generally speaking, dedicating the net proceeds to drug treatment;

(5) Subsection (8), prohibiting the use for law enforcement purposes of federal civil forfeiture proceeds that are shared with state and local law enforcement agencies;

(6) Subsection (9), prohibiting the transfer of a civil forfeiture proceeding to the federal government unless a state court has found that the “activity giving rise to the forfeiture is interstate in nature and sufficiently complex,” that “[t]he seized property may only be forfeited under federal law,” or that pursuing the forfeiture under state law “would unduly burden” the state forfeiting agency; and

(7) Subsection (11), establishing a reporting requirement for all civil forfeitures.

Plaintiffs further assert that subsection (3) of Measure 3 implicitly changed Article I, section 16, of the Oregon Constitution, which prohibits the imposition of “excessive fines” in criminal proceedings. Plaintiffs contend that, for purposes of Article I, section 16, a civil forfeiture action is a criminal proceeding. They argue that, by imposing the requirement that a forfeiture be “substantially proportional to the specific conduct,” subsection (3) changed the standard of Article I, section 16, which, they assert, merely required that a fine not be “grossly disproportionate” to the prohibited conduct on which the fine is based. Plaintiff intervenors assert, in addition, that Measure 3 “explicitly links forfeitures to criminal law standards and convictions” and changes Article I, section 16, by changing the meaning of “offense” or adding a requirement that forfeiture not take place unless *535the property is found to have been instrumental in the crime or the proceeds of the crime.

In sum, then, plaintiffs assert that Measure 3 made at least seven changes by addition to the Oregon Constitution and that it also implicitly made at least one change to an existing provision of the constitution. Plaintiffs also argue that those changes were substantive changes to the constitution that were not “closely related” and, thus, that Measure 3 denied the voters the opportunity to vote separately on each unrelated substantive change, in violation of Article XVII, section 1, as the Oregon Supreme Court interpreted that section in Armatta v. Kitzhaber, 327 Or 250, 959 P2d 49 (1998).

Defendants and defendant intervenors respond that Measure 3 made only one substantive change to the Oregon Constitution and that, even if it made two or more substantive changes, the changes were “closely related” and, therefore, Measure 3 did not violate the “separate-vote” requirement of Article XVII, section 1..

In Armatta, the Supreme Court first determined that the separate-vote requirement applies to constitutional amendments proposed by initiative, as well as to those proposed by the legislature. 327 Or at 261. Moreover, the court concluded that the concept of a constitutional “amendment” meant “a particular constitutional change,” id. at 266-67, and, thus, it held that the separate-vote requirement requires that proposed amendments to the constitution be submitted to the voters in a manner that permits the voters to “express their will in one vote as to only one constitutional change.” Id. at 269. The court then discussed and analyzed the specific wording of the section, its historical development, and case law applying it. Id. at 269-75.

From those analytical tools, the court derived the following test for determining whether a particular proposal to amend the constitution offends the separate-vote requirement:

“[T]he proper inquiry is to determine whether, if adopted, the proposal would make two or more changes to the constitution that are substantive and that are not closely related. If the proposal would effect two or more changes that are *536substantive and not closely related, the proposal violates the separate-vote requirement of Article XVII, section 1, because it would prevent the voters from expressing their opinions as to each proposed change separately.”

Armatta, 327 Or at 277.

In applying Armatta, we look not only at explicit changes but also at the implicit changes, if any, that a measure would make to the constitution to determine whether it offends the separate-vote requirement of Article XVII, section 1. See id. at 278 (identifying explicit and implicit changes). Moreover, constitutional changes are not limited to amendments that expressly or implicitly repeal or modify existing constitutional provisions; additions to the constitution also can make substantive changes to it. See Lehman v. Bradbury, 333 Or 231, 243, 37 P3d 989 (2002) (by adding Article II, section 19, to the constitution, Ballot Measure 3 (1992) changed the constitution by limiting state legislative service to a lifetime maximum of six years in the Oregon House of Representatives, even though no provision of the existing constitution had addressed the matter of congressional term limits).

As noted, Measure 3 did not expressly repeal or modify any provision of the existing constitution, but the parties vigorously disagree as to whether Measure 3 implicitly changed various existing constitutional provisions. Accordingly, we could begin our separate-vote inquiry by focusing on that issue. See Armatta, 327 Or at 278 (beginning inquiry concerning ballot measure by identifying changes that it would have made to existing constitution). However, the Supreme Court has not adopted a fixed order of analysis for determining whether an amendment violates the separate-vote requirement. See Swett v. Bradbury, 333 Or 597, 607, 43 P3d 1094 (2002) (stating that the order of analysis followed in Lehman was “descriptive, not prescriptive”). We need not resolve the parties’ disagreement concerning implicit changes if Measure 3 made at least two substantive changes by addition to the existing constitution that are not closely related. As we now explain, Measure 3 made at least two substantive changes by addition to the existing constitution.6 If *537those changes are not closely related, the entire measure fails to pass muster under the separate-vote requirement. Accordingly, we first consider those changes and whether they are closely related.

Measure 3 added provisions to Article XV that, among other effects, prohibit the civil forfeiture of property unless its owner first has been convicted of a crime (subsection 3) and that direct how forfeited property is to be disposed of and distributed (subsection 7). Plaintiffs argue that those provisions make discrete substantive changes to the constitution. For two reasons, defendants disagree.

First, they argue that Measure 3 made a single change, albeit through multiple subsections, that created a “constitutional limitation on forfeiture of property.” That argument is unpersuasive. Broadly viewed, Measure 3 may address a single subject that encompasses multiple aspects of civil forfeiture reform. However, in Lehman, the Supreme Court summarily rejected the notion, advanced by the state there, that Measure 3 (1992) made only one change to the constitution because it was a “cohesive combination of provisions” on a “discrete subject,” i.e., term limits. Id. at 244. The court explained:

“[T]o the extent that intervenors are arguing that Measure 3 does not make multiple changes in the ordinary sense of multiple substantive alterations to existing wording, their argument is unsupportable. Changes in the term limits for state executive officers and the creation of such limits for state legislators and for members of Congress are at least two substantive changes to the constitution.”

Id. Defendants’ first argument reduces to the same notion rejected in Lehman, that is, that any number of alterations to the wording of the constitution constitute a single change if they reflect or implement a single cohesive policy goal, in this case, restricting civil forfeiture proceedings. Whatever the *538abstract appeal of such an argument, it is plainly foreclosed by Lehman.

Defendants’ second argument is equally unavailing. Defendants assert that, because subsections 3 and 7 of Measure 3 do not expressly or implicitly modify any existing provision of the constitution, they “could be viewed as a single change to the constitution.” However, the Supreme Court has not adopted a different meaning of “change” depending on whether new provisions modify or repeal existing constitutional provisions. The court was presented with that opportunity in Lehman when it was required to determine whether the addition of a new provision, pertaining to congressional term limits, constituted a discrete substantive change to the constitution. In treating that change as distinct from the change to existing provisions pertaining to term limits for state officials, the court did not suggest that it was applying different standards for determining whether a change had been made. Nor do we.

We conclude that, in requiring a criminal conviction as a condition to civil forfeiture and, further, in imposing restrictions on the use of forfeiture proceeds, Measure 3 made at least two substantive changes to Article XV.7

We turn to the remaining question under Armatta, namely, whether the substantive changes to the constitution made by subsections 3 and 7 of Measure 3 are “closely related.” In aid of a proper application of that test, we begin by reviewing the court’s cases applying it.

In Armatta, the court did not elaborate on its application of the “closely related” test to the substantive changes made by the amendment at issue, Ballot Measure 40 (1996) (Measure 40). That was so because, once the court had described the multiple, substantive, changes that Measure 40 made to the constitution, it concluded that “their lack of ‘close relation’ was obvious.” Lehman, 333 Or at 244 (so describing Armatta). Among other changes, Measure 40 *539(1) limited the right of all persons to be free from unreasonable searches and seizures, self-incrimination, and former jeopardy under Article I, sections 9 and 12, by granting crime victims the right to have “all relevant evidence admissible against the criminal defendant”; (2) limited the right of a defendant in a murder case, under Article I, section 11, to a unanimous verdict by permitting convictions for murder based on a jury’s 11-1 “guilty” vote; and (3) limited the legislature’s authority under Article VII (Amended), section 5(l)(a), to enact laws pertaining to juror qualifications by requiring that jurors in criminal cases be registered voters and not have been convicted of a felony within the previous 15 years. Armatta, 327 Or at 254.8 Under those circumstances, the court concluded, without extended discussion, that the changes made by Measure 40 were not “closely related” and, thus, that Measure 40 violated the separate-vote requirement of Article XVII, section 1. Id. at 283.

Armatta nevertheless provides some guidance for resolving the question whether the changes that Measure 3 made to the Oregon Constitution were closely related. First, the court noted that at least two constitutional provisions affected by Measure 40—Article I, section 9, and Article I, section 11—conferred rights that had “virtually nothing to do” with each other, that is, the right of all people to be free from unreasonable searches and seizures and the right of the criminally accused to have a unanimous verdict rendered in a murder case. Armatta, 327 Or at 283. The court further distinguished those two constitutional provisions by the fact that each involves “separate constitutional rights, granted to different groups of persons.” Id. For the voters to change such provisions in a substantive way, the court held, they must vote on the changes separately. Id. at 284; see also Lehman, 333 Or at 246 n 9 (“when the affected constitutional provisions confer separate rights on different groups of people, that is a strong indication that those provisions are not ‘related’ for purposes of the separate-vote requirement”).

*540In addition, the court suggested in Armatta that Measure 40 had a second weakness, in that the constitutional changes that it made were very different from one another. For example, one change made by Measure 40— adding the requirement that the jury pool in criminal cases be drawn from registered voters—did not “share [the] relationship” that bound certain other changes to each other. Id. at 283. In making that observation, the court, in effect, considered both the relationship among the constitutional provisions affected by Measure 40 and the relationship of the constitutional changes that were made in those provisions to one another.

Lehman also guides our analysis. In Lehman, the challenging plaintiffs argued that the substantive change to the Oregon Constitution made by the addition of a provision that limited the terms of members of Congress was not closely related to the substantive changes that limited the terms of state officials. 333 Or at 248. The court agreed. It reasoned that, although both provisions created term limits for elected officials, nevertheless, in doing so, they made unrelated changes to the constitution. Id. at 249. The court explained:

“When the people were asked in Measure 3 to add section 20 to Article II of the Oregon Constitution, they were asked to change the eligibility of members of Congress, a topic found nowhere else in the Oregon Constitution. The problem was not necessarily that the provision was new. Newness, in and of itself, may be a neutral factor. But the specific addition made by section 20, affecting eligibility for federal public office, had little or nothing to do with term limits for the Oregon State Treasurer, for example, as those limits were established in section 19. Nonetheless, the voters were asked to vote for or against both sections in a single measure. In terms of Article XVII, section 1, Measure 3 submitted two or more amendments to the voters in a manner that prevented the voters from voting on each amendment separately. That was impermissible.”

Lehman, 333 Or at 250.

Swett also is instructive. There, the court considered the constitutionality, under Article XVII, section 1, of Ballot Measure 62 (1998), which established various requirements *541for, among other things, disclosure of political contributions and initiative petition signature gathering. The court concluded that, at a minimum, section 1 of Measure 62 implicitly changed Article II of the constitution, which contains various eligibility requirements for publicly elected officials, by imposing an additional requirement on those publicly elected officials if they received political contributions in excess of $500 from any one contributor, Swett, 330 Or at 607-08, the officials would have been required to disclose the amount and source of the contribution. Id. at 603. Meanwhile, section 3 of Measure 62 affected Article IV, section 1(2), concerning the legislative power: It explicitly amended that constitutional provision to require that persons who gather signatures for initiative and referendum petitions be registered voters. Id. Thus, the measure made more than one substantive change to the constitution.

The defendants in Swett contended that the constitutional changes made in sections 1 and 3 were closely related because, like other changes proposed by Measure 62, “ ‘they are regulations designed to prevent, control, or expose the influence of money in the initiative [and] referendum * * * process.’ ” 333 Or at 608 (brackets in Swett). However, the court held that the voter-registration requirement of section 3 did not bear any relationship to that suggested design. Id. at 609.

More importantly, the court concluded that the defendants’ argument failed because it was merely an attempt to show that sections 1 and 3 of Measure 62 shared the same subject matter. The court said:

“That may or may not be true, but it is beside the point in an analysis under Article XVII, section 1. Defendants do not focus, as they must in a separate-vote challenge, on the particular changes made to the constitution. See Lehman[, 333 Or] at 241-42 (separate-vote requirement, in contrast to single-subject requirement, focuses on extent to which proposed amendment modifies existing constitution). Section 1 of Measure 62, by imposing a disclosure requirement on those who receive political contributions, adds an additional rule to those already set out in Article II concerning how a publicly elected official must act. By contrast, section 3 of Measure 62, by creating an eligibility requirement for *542those who undertake the task of gathering signatures for initiative and referendum petitions, limits the legislative power of the people as described in Article IV, section 1(2). Those two constitutional changes, as far as we can tell either from our own analysis or on the basis of defendants’ arguments, are not closely related to each other.”

Swett, 333 Or at 609 (emphasis in original). Accordingly, the court invalidated the measure.

The court most recently considered whether constitutional changes were closely related for purposes of the separate-vote requirement in League of Oregon Cities v. State of Oregon, 334 Or 645, 56 P3d 892 (2002). In that case, the plaintiffs challenged the constitutionality of Ballot Measure 7 (2000). The court held that Measure 7 made at least two substantive changes to the existing constitution, namely, a change to the requirements for payment of just compensation to property owners under Article I, section 18, and a separate change respecting laws restraining expression under Article I, section 8. Id. at 673-74. The court held:

“Those two constitutional provisions involve ‘separate constitutional rights, granted to different groups of persons.’ Armatta, 327 Or at 283. That lack of relationship ‘is a strong indication that those provisions are not “related” for purposes of the separate-vote requirement.’ Lehman, 333 Or at 246 n 9.”

League of Oregon Cities, 334 Or at 674-75.

Turning to the constitutional changes themselves, the court concluded that the change that Measure 7 made to Article I, section 18, was not closely related to the change that it made to Article I, section 8. Id. at 675. The change to Article I, section 18, generally concerned the ability of a real property owner to obtain just compensation for enforcement of certain regulations. By contrast, the change to Article I, section 8, operated to limit the rights of certain property owners vis-á-vis other property owners, based on the content of the expressive material sold on their property. The court concluded that, “[b]y incorporating both types of substantive changes that are not closely related, Measure 7 makes ‘two or more amendments’ to the Oregon Constitution that the voters were required to have voted upon separately.” Id.

*543We discern several core principles from the foregoing decisions that inform our decision here. First, when an amendment makes substantive changes by addition to the existing constitution, we look at the extent to which the changes themselves are closely related. Lehman, 333 Or at 243, 250. The fact that the new changes do not alter any provision of the existing constitution does not appear to subject them to a more lenient standard for purposes of determining whether they are closely related to each other. See id.

Second, we must not confuse the single-subject and separate-vote requirements. Although interrelated, they are distinct constitutional principles. Swett, 333 Or at 609; Lehman, 333 Or at 241-42; Armatta, 327 Or at 256-75. Irrespective of whether two or more constitutional changes share a common subject or objective, they still must be closely related to each other in order to satisfy Article XVII, section 1. In effect, that determination is concerned with the strength of the logical relationship between the changes. Swett, 333 Or at 609; Lehman, 333 Or at 250; Armatta, 327 Or at 283. In that regard, in considering whether the changes are closely related, it is useful to identify the persons or groups of persons, if any, whose rights the changes affect. If the changes involve “separate constitutional rights, granted to different groups of persons,” there is a “strong indication” that the changes are not closely related for purposes of the separate-vote requirement. League of Oregon Cities, 334 Or at 674; Lehman-, 333 Or at 246 n 9; Armatta, 327 Or at 283.

Defendants and defendant intervenors rely on two Supreme Court decisions, which, they assert, follow a different set of principles for purposes of determining whether constitutional changes are closely related. The first is Baum v. Newbry et al., 200 Or 576, 267 P2d 220 (1954). In Baum, the plaintiff argued that a 1952 amendment to Article IV, section 6, of the Oregon Constitution concerning reapportionment violated the separate-vote requirement. In rejecting that challenge, the court upheld a measure that modified and, indeed, eliminated some existing constitutional requirements and added new requirements.

The challenged initiative measure altered the existing constitution in several respects. First, it changed existing *544Article IV, section 6, to eliminate the state’s authority to conduct its own census or to rely exclusively on the “number of white population” in making reapportionment calculations. After the 1952 amendment, only the United States census could be used and nonwhites would be counted. The then-existing reapportionment formula also was changed. The measure also added new subsections to Article IV, section 6. Two of those new provisions vested original jurisdiction in the Supreme Court to review reapportionment plans prepared by either the legislature or the Secretary of State and authorized the Secretary of State to enact reapportionment laws if the legislature failed to do so. Two other new provisions added to the constitution a temporaiy but detailed reapportionment plan, specifying the number of senators and representatives for each district and county or counties. Another new provision added rules governing the service of senators who would be elected from existing districts but who would finish their terms under the new reapportionment plan. Finally, another new provision specified that the measure generally did not become effective until the day of the 1954 election.

The Supreme Court rejected the plaintiffs separate-vote challenge, stating:

“But the 1952 constitutional amendment did not submit ‘two or more amendments’ to the voters. It submitted one amendment which deals only with the subject of reapportionment of the members of the legislative assembly and with matters which are germane thereto. * * * Section 1 of Article XVII does not prohibit the people from adopting an amendment which would affect more than one article or section by implication. Annotation, 94 ALR 1510. At most it prohibits the submission of two amendments on two different subjects in such manner as to make it impossible for the voters to express their will as to each. The fact, if it be one, that the reapportionment amendment may have amended more than one section of the constitution, would be immaterial.”

Baum, 200 Or at 581. In Armatta, the court said that Baum stands for the following principles:

*545“First, it demonstrates that the purpose of the separate-vote requirement is to allow the voters to decide upon separate constitutional changes separately. Stated differently, Article XVII, section 1, imposes a requirement aimed at ensuring that the voters are able to express their will in one vote as to only one constitutional change. That is consistent with our textual analysis of the separate-vote requirement, which noted that the requirement focuses upon the nature of the change to the existing constitution, as well as the procedural form that an amendment takes when it is submitted to the people. Second, Baum demonstrates that, by implication, a single constitutional amendment may affect one or more constitutional provisions without offending the separate-vote requirement. Finally, Baum suggests that the separate-vote requirement encompasses, to some extent, the notion that a single amendment must contain a single ‘subject.’ ”

327 Or at 269.

In the second case on which defendants and defendant intervenors rely, Hartung v. Bradbury, 332 Or 570, 579-80, 33 P3d 972 (2001), the court reaffirmed Baum as having correctly applied the separate-vote requirement. In Hartung, the court rejected a separate-vote challenge to a 1986 amendment to Article VI, section 6, that repealed the 1952 version and replaced it with a new version that excepted reapportionments from the initiative and referendum powers and altered the procedure for recall of legislators in reapportioned districts. The court said that “nothing in Armatta suggests that Baum was decided incorrectly[.]” 332 Or at 579 n 5. It held that, “[i]n light of this court’s conclusion in Baum that the more extensive 1952 amendment passed muster under Article XVII, section 1, we conclude that the more limited 1986 amendment necessarily withstands petitioners’ constitutional challenge.” Id. at 579.

Defendants and defendant intervenors argue that the constitutional changes made by Measure 3 are at least as closely related as those challenged in Baum. Defendants assert:

“[In Baum], the court upheld a measure that altered the general rules for apportioning members of the legislature, inserted the Secretary of State into the reapportionment *546process, established a temporary reapportionment, and eliminated the requirement that only ‘white’ persons be counted in making the reapportionment calculations. The Baum court did not clearly explain its conclusion that the measure satisfied the separate-vote requirement because it found that these were not multiple substantive changes or because, if they were, they were closely related. In any event, the ‘changes’ in Measure 3 cannot be said to be less closely related than the reapportionment changes. All of the changes in the reapportionment measure were directed to establishing a new mechanism for accomplishing that task. Similarly, all of the changes in Measure 3 are directed to establishing protections for private property owners faced with civil forfeiture proceedings.”

Defendants and defendant intervenors assert that it is difficult to reconcile the court’s decisions in Baum and Hartung with its reasoning in Armatta, Lehman, Swett, and League of Oregon Cities. That proposition is, at least, debatable. For two reasons, Baum, in particular, presents an analytical challenge. First, that decision appears to blur the line between the single-subject and separate-vote requirements that the court has, in later decisions, attempted to draw. AlS noted, in Baum, the court held that, “[a]t most [Article XVII, section 1,] prohibits the submission of two amendments on two different subjects in such manner as to make it impossible for the voters to express their will as to each.” Baum, 200 Or at 581 (emphasis added). The constitutional changes made by the measure challenged in Baum appear to have been connected only by a broad unifying subject matter, that is, reapportionment. Second, those constitutional changes arguably conferred rights on at least two different groups of people, namely, persons of color and persons who, regardless of race, are residents of the various geographical areas affected by the redistricting inherent in reapportionment.

That arguable tension notwithstanding, it is apparent to us that the controlling principles for determining whether constitutional changes are closely related for purposes of Article XVII, section I, are those identified in Armatta and its progeny. We now apply those principles to subsections 3 and 7 of Measure 3. Because those provisions *547do not repeal or modify provisions of the existing constitution, we proceed directly to determine whether the provisions themselves are closely related.9

We hesitate to say that there is no relationship between the two provisions because each, in its own way, arguably is aimed at making forfeiture more difficult to accomplish. Subsection 3 imposes an obstacle to forfeiture in the form of a condition precedent. Subsection 7, in turn, creates a possible disincentive to forfeiture proceedings by prohibiting the use of forfeiture proceeds for the law enforcement purposes for which they historically were used, directing them instead to drug treatment.

Nevertheless, that common objective does not make the changes themselves closely related. Swett, 333 Or at 609; Lehman, 333 Or at 250; Armatta, 327 Or at 283. If the constitutional changes at issue in Lehman, both of which related to a specific objective—the creation of term limits for public officials—had “little or nothing to do” with each other, Lehman, 333 Or at 250, then it is difficult to conceive how the constitutional changes effected by subsections 3 and 7 of Measure 3 could be regarded as closely related. There is little, if any, logical relationship between subsection 3, a provision that gives a property owner the right to be convicted of a crime before his or her property may be forfeited, and subsection 7, which directs and restricts the purposes for which forfeiture proceeds can lawfully be used.

In determining whether the provisions are closely related, we also consider whether they affect the rights of different groups of people. Here, they arguably do. Subsection 3 *548confers a right on owners of property subject to possible forfeiture, for whom a criminal conviction must precede forfeiture. Subsection 7, on the other hand, arguably confers a benefit on a different group, persons receiving drug treatment.10

For the foregoing reasons, we conclude that subsections 3 and 7 of Measure 3 made two substantive changes to the Oregon Constitution that are not closely related.11

The dissent disagrees with our conclusion. The heart of that disagreement concerns the relationship between the separate-vote and single-subject requirements. The dissent says:

“[T]he separate-vote requirement is concerned with the structure of state government and the structure of the constitution. Whether two changes are closely related for purposes of that requirement concerns how the changes fit together when viewed in terms of governmental and constitutional structure. The focus is not on whether the changes achieve policy goals that are closely related to each other.”

*549188 Or App at 568 (Armstrong, J., dissenting) (emphasis in original). The problem is that the differences between the two constitutional requirements are not those that the dissent identifies. Armatta makes the point this way:

“[W]e acknowledge that, under Baum, 200 Or at 581, the separate-vote requirement encompasses the notion that a single constitutional amendment must contain what the court there referred to as a single ‘subject[ ].’ Indeed, if a proposed amendment contained two different subjects, then it could not be considered a single amendment, regardless of the existence of the single-subject requirement of Article IV, section l(2)(d). However, the fact that a proposed amendment containing more than one subject would violate both the separate-vote and single-subject requirements does not compel the conclusion that the opposite also is true, i.e., that a proposed amendment that contains only one subject would not violate the separate-vote requirement. As we have discussed, the separate-vote requirement imposes a narrower restriction than the requirement that a proposed amendment embrace only one subject. It follows, therefore, that a proposed amendment that satisfies the broad standard for embracing a single subject nonetheless may violate the separate-vote requirement.”

Armatta, 327 Or at 276-77. Because the separate-vote requirement encompasses the requirement that a single constitutional amendment contain a single subject, a proposed amendment that would violate the single-subject requirement necessarily would violate the separate-vote requirement. Id. Accordingly, contrary to the dissent’s view, the separate-vote requirement is concerned with the subject matter of multiple changes.12

Of course, the separate-vote requirement is concerned with more than subject-matter identity. But, in trying to describe the essence of that additional ingredient, the dissent becomes confusing and unhelpful. For example, the dissent asserts that, if all of the provisions of an amendment *550could have been included as a “single component in the relevant constitutions, given the manner in which those constitutions have been divided into discrete components,” 188 Or App at 557 (Armstrong, J., dissenting), they will pass muster under Article I, section 17. The dissent apparently believes that, in deciding whether multiple constitutional changes violate Article XVII, section 1, it is important, if not essential, to consider the manner in which constitutions have “traditionally’ been structured. Id.

We are skeptical of the validity of that proposition as a touchstone for determining close relatedness. Although the structure of the Oregon Constitution may provide clues as to whether constitutional changes are closely related to existing provisions of the constitution, it should not be a decisive factor in that determination. Many “discrete components” of the constitution exist in the form they do for historical, as opposed to logical, reasons. Moreover, those forms often involve the layering of amendments that have been added to the constitution over the years. Thus, provisions in completely different articles of the constitution could be more closely related to each other, from a logical standpoint, than provisions that occupy the same article or even the same section.13 The structure of the existing constitution, then, is not a wholly reliable surrogate for determining the close relatedness of constitutional changes.

The problem with the dissent’s theory is most apparent in the context of its attempt to explain Lehman. The dissent states that “the qualifications of elected state officials present very different issues in terms of the structure of state government and of the state constitution from those of the members of Oregon’s congressional delegation.” 188 Or App at 568 (Armstrong, J., dissenting). However, the dissent fails to adequately explain why the changes effected by subsections 3 and 7 of Measure 3 are more closely related to each other than those at issue in Lehman. To its credit, the dissent strives valiantly to distinguish Lehman. It points out that the *551United States Supreme Court held in US Term Limits, Inc. v. Thornton, 514 US 779, 115 S Ct 1842, 131 L Ed 2d 881 (1995), that a provision imposing term limits on members of Congress cannot properly be made part of a state constitution. According to the dissent, that vice in Measure 3 (1992)

“serves to emphasize the point that I have tried to make— that state policies affecting state officials present very different issues than do state policies affecting federal officials. The state and federal officials are chosen by the same electors, but they perform functions for separate sovereigns. Even if a state constitution could be amended to include a provision on the qualification of members of Congress, the amendment that made that change would have to be adopted separately from one addressed to the qualifications of state elected officials because, as a matter of constitutional structure, state and federal officials are fundamentally distinct. Although the policy goal and subject of the two amendments would be the same—limiting the term of office of officials elected by Oregon electors—the state officials affected by the policy would be distinct from the federal officials affected by the policy when viewed in terms of the structure of the state and federal governments and, consequently, the structure of the two constitutions.”

188 Or App at 563-64 (Armstrong, J., dissenting) (emphasis in original). There are two problems with that argument.

First, the unlawfulness under the United States Constitution of the congressional term limits provision in Measure 3 (1992) had nothing to do with the court’s determination in Lehman that it was not closely related to the provision imposing term limits for state officials. See Lehman, 333 Or at 235 (“the fact that section 20 violates the United States Constitution plays no role in our analysis concerning whether its presence in Measure 3 causes that measure to violate Article XVII, section 1, of the Oregon Constitution”). Thus, the fact that the voters could not lawfully make one of the changes was not the problem for purposes of Article XVII, section 1.

Second, returning to our broader concern, the dissent’s focus is misplaced when it says that the state officials affected by the policy are not “distinct from,” 188 Or App at 564 (Armstrong, J., dissenting), the affected federal officials *552“when viewed in terms of the structure of the state and federal governments and, consequently, the structure of the two constitutions.” Id. (Armstrong, J., dissenting). Close relatedness depends only in part on the relationship between changes to the constitution and existing provisions of the constitution. Regardless of their relationship to existing constitutional provisions, the changes themselves must, in order to satisfy Article XVII, section 1, be closely related to each other.

Lehman clearly reinforces that point. Before Measure 3 (1992) was adopted, there were no term limit provisions in the Oregon Constitution for either state or federal officials. Although, when Measure 3 (1992) was adopted, there were provisions in the existing Oregon Constitution that related to state officials in other respects, the court expressly declined to consider whether the change made by the term limits provision for state officials was closely related to those preexisting provisions. Lehman, 333 Or at 248. Instead, the close relatedness of the changes was determined solely by comparing them to each other. Id. at 249. Accordingly, the dissent’s theory of close relatedness does not explain the result in Lehman, and it does not weaken our conclusion that the changes effected by subsections 3 and 7 of Measure 3 are not closely related to each other.

The dissent also rows into other unnavigable theoretical waters. Despite its ultimate conclusion that the separate-vote requirement does not focus on the relationship between the policy goals of multiple amendments, it initially suggests that the changes made by Measure 3 are closely related because it was “intended to establish a coherent and comprehensive scheme to address a discrete policy goal.” 188 Or App at 565 (Armstrong, J., dissenting). To its credit, the dissent later acknowledges that “it is not enough to show that the provisions of an amendment serve such a purpose.” 188 Or App at 566 (Armstrong, J., dissenting). In fact, the Supreme Court rejected that test for close relatedness in Lehman, 333 Or at 242.

At another point, though, the dissent explains, in still different terms, why it believes that the changes made by Measure 3 are closely related. It says that Measure 3 *553“embodies a series of policy decisions about civil forfeiture, but those policy decisions are ones that had to be addressed in order to establish a comprehensive system of civil forfeiture.” 188 Or App at 566 (Armstrong, J., dissenting) (emphasis added). That statement embellishes the cohesive policy choice theme by adding a dimension of necessity. Not only does it attempt to promote a failed restatement of the close relatedness test; it is also factually inaccurate. Measure 3 did not purport to establish a comprehensive system of civil forfeiture; instead, it selectively targeted various components of the system in an apparent effort to make forfeiture more difficult to accomplish.

In the end, the dissent’s analysis is a collage of flawed restatements of the requirement that multiple constitutional changes be closely related in order to pass scrutiny under Article XVII, section l.14 Although they are well-intentioned, we cannot endorse them.

As the court explained in Armatta, a proposed constitutional amendment must be adopted in compliance with constitutional requirements:

“It is a long-standing principle of law that a proposed constitutional amendment must be adopted in compliance with the procedures set forth in the Oregon Constitution:

“ ‘The provisions of the constitution for its own amendment are mandatory, and must be strictly observed. A failure in this respect will be fatal to a proposed amendment, notwithstanding it may have been submitted to and ratified and approved by the people. The constitutional provisions are as binding upon the people as upon the legislative assembly, and the people cannot give legal effect to an amendment which was submitted in disregard of the limitations imposed by the constitution * * *.’ ”

*554Armatta, 327 Or at 284 (quoting Kadderly v. Portland, 44 Or 118, 135-36, 74 P 710 (1903) (emphasis in Armatta)). Accordingly, because Measure 3 was not adopted in compliance with the requirements of Article XVII, section 1, we hold that it is void in its entirety.15

Reversed.

dissenting.

The majority holds that Ballot Measure 3 (2000) violated Article XVII, section 1, of the Oregon Constitution by making two or more amendments to the constitution without submitting each amendment to the voters for a separate vote.1 Although I agree with the majority that Measure 3 made several substantive changes to the constitution, I disagree with its conclusion that at least two of the changes had to be adopted separately because they were not related closely enough to be adopted in a single amendment. I conclude that they were closely related and, consequently, that they did not have to be voted on separately. I also conclude that the measure complied with the single-subject requirement in Article IV, section l(2)(d).2 I therefore would affirm the trial court’s judgment that declared that Measure 3 did not violate either constitutional provision.

I commend the majority for its effort to make sense of the principle in Article XVII, section 1, that requires separate votes on separate amendments to the constitution. The text of the provision provides little guidance about its meaning, and the cases interpreting it provide little direct analytical help either. I am convinced, however, that the majority misunderstands the provision and the cases that have applied it.

*555Article XVII, section 1, requires that separate amendments to the Oregon Constitution be voted on separately. Equivalent requirements are found in many state constitutions.3 The provision complements Article XVII, section 2, which governs the process by which the constitution can be revised. Viewed together, the provisions establish a comprehensive system for altering the constitution.

An amendment involves a discrete, circumscribed change to the constitution. It can be proposed by initiative petition or by legislative referral. A majority vote by both houses of the legislature is required for the legislature to refer an amendment to the voters for their approval. Or Const, Art XVII, § 1.

In contrast, a revision involves a constitutional change that affects more than a discrete, circumscribed part of the constitution. It can be proposed only by the legislature, and it requires a two-thirds vote by both houses for it to be referred to the voters for approval. Id. at § 2.

For the amendment and revision provisions to work, they necessarily require that there be a standard or test by which to determine what constitutes an amendment to the constitution. Surprisingly, Armatta v. Kitzhaber, 327 Or 250, 959 P2d 49 (1998), was the first Oregon case to establish such a standard and one of the first in the country to do so. It is noteworthy that courts in several states have relied on Armatta in interpreting comparable provisions in their state constitutions.4

I will discuss Armatta in greater detail later in this opinion. What is important to note here is that it established a two-part test for determining whether a proposed constitutional change constitutes an amendment. Under the test, people who draft amendments and courts that enforce the separate-vote requirement for amendments first must determine whether a proposed amendment makes more than one *556substantive change to the constitution. If it does, then drafters and courts must determine whether the substantive changes are closely related. If they are, then the proposed amendment can properly be adopted as an amendment. Conversely, if the substantive changes are not closely related, then the proposed amendment cannot properly be adopted as an amendment. See id. at 275-77.

Armatta held that Ballot Measure 40 (1996) violated the test that the court established for the separate-vote requirement by making two or more substantive changes to the constitution that were not closely related. 327 Or at 283-85. Although the court explained in Armatta why the substantive changes on which it focused in the case were not closely related, see id., the court did not attempt to describe or further define the characteristics of substantive changes that would be closely related enough to meet the constitutional test.

The court has since refused to refine or alter the test established in Armatta, see Lehman v. Bradbury, 333 Or 231, 240-42, 37 P3d 989 (2002), which makes sense. Although the Armatta test is imprecise, it may be the best test available to implement the separate-vote requirement in Article XVII, section 1, given the broad range of conceivable substantive changes that could be made to the constitution. That means that court decisions that correctly apply the test or that are consistent with it will provide important guidance about the test.

A review of the Oregon cases that have applied the separate-vote requirement suggests that the structure of the constitution, itself, bears on whether a proposed amendment complies with the requirement. The Oregon Constitution and other comparable constitutions are structured as they are because of judgments that their drafters made about how different, discrete policies should be embodied in them. Some of the divisions within them are arbitrary, in that the constitutions could have been sensibly designed with fewer or different subdivisions.5 Nevertheless, the way in which the *557relevant constitutions are structured bears on whether a proposed amendment constitutes a single amendment or multiple amendments. A proposed amendment satisfies the separate-vote requirement if it addresses a discrete subject in a way that is consistent with the manner in which comparable subjects have been segregated into discrete components in the constitutions. That means that the provisions of a proposed amendment bear a close enough relationship to each other to be part of a single amendment if the provisions, when considered together, could reasonably have been included as a single component in the relevant constitutions, given the manner in which those constitutions have been divided into discrete components.6

As I will explain, that understanding of the separate-vote requirement is consistent with and, I believe, implicit in the Supreme Court cases that have applied -it. However, the understanding that I have described can perhaps best be explored through consideration of several hypothetical amendments to the constitution.

Assume, for example, that the Oregon Constitution did not have a Bill of Rights at its adoption and that a proposed amendment sought to add one to it. The proposed amendment arguably could satisfy the single-subject requirement in Article IV, section l(2)(d), of the constitution because it would seek to add to the constitution a body of civil liberties guarantees against the government. However, those guarantees traditionally have been embodied in separate provisions that address discrete subjects, such as free speech, religious liberty, the right to bear arms, etc. Consequently, as a matter of constitutional structure, the separate-vote requirement would require the different guarantees to be *558submitted as separate amendments, as was done with the federal Bill of Rights.

The same principle would apply to the particular guarantees themselves. The First Amendment to the United States Constitution contains several guarantees: religious liberty, freedom of speech and of the press, and freedom of assembly. US Const, Amend I. In contrast, the Oregon Bill of Rights contains six separate provisions on religious liberty, a separate provision on freedom of speech, and a separate provision on freedom of assembly. See Or Const, Art I, §§ 2-8, 26. The decision by the drafters of the Oregon Constitution to separate the religious liberty, speech, and assembly provisions in the manner that they did arguably suggests that each provision was understood to be independently significant, which suggests that the separate-vote requirement might require that each provision be amended by a separate amendment. However, the First Amendment provides some support for the idea that the various guarantees could sensibly be addressed in a single amendment. Furthermore, a good argument could be made that, notwithstanding the existing provisions of the Oregon Bill of Rights, at least some of the religious liberty guarantees are closely related enough so that they could be adopted or amended by a single amendment and not run afoul of the separate-vote requirement. See id. at §§ 2-4. The same point could be made about the free speech and free assembly provisions. See id. at §§ 8, 26.

In the same vein, it might not be possible to adopt or amend the constitution’s legislative article, Article IV, in a single amendment without violating the separate-vote requirement. It might be possible, however, to address portions of that article, such as the provisions on the form and adoption of bills, in a single amendment without violating the requirement. That might be so because, while there are several, separate provisions that address that subject in the constitution, see id. at Art IV, §§ 18-22, the different aspects of the subject that they address are closely related enough for them to be sensibly combined in a single amendment without conflicting with the structural principle embodied in the separate-vote requirement.

*559The Oregon cases are consistent with that understanding of the separate-vote requirement. Baum v. Newbry, 200 Or 576, 267 P2d 220 (1954), the first case that applied the requirement, provides a useful starting point from which to analyze the issue. Baum involved a separate-vote challenge to a 1952 amendment to Article IV, section 6, of the Oregon Constitution, the provision that governs the reapportionment of the Oregon legislature. The amendment made a number of substantive changes to the constitution. They included changes in the census used to apportion the legislature, creation of direct Supreme Court review of reapportionment plans, imposition of a requirement that the Secretary of State reapportion the legislature if the legislature fails to do so, alteration of the formula by which senators and representatives are to be reapportioned, and adoption of a reapportionment plan for the legislature that took effect in 1954. Or Const, Art IV, § 6 (1952). Although the amendment made those changes, the court rejected the plaintiffs contention that Article XVII, section 1, required a separate vote on any of them. The court was right to do that.

The 1859 constitution required the legislature to reapportion itself at least every ten years. Or Const, Art IV, § 6 (1859). However, as of 1952, the legislature had failed for 45 years to reapportion itself, and there was no apparent mechanism by which the requirement could be enforced. The people who proposed the 1952 amendment sought to correct the problem. Although the goal of obtaining valid and timely reapportionments of the legislature could have been accomplished in a variety of ways, all of the provisions of the 1952 amendment represented sensible choices to achieve that singular goal.

More fundamentally, however, the 1859 constitution addressed legislative reapportionment in a single section. See id. Although the 1952 amendment established a more complex and comprehensive scheme to ensure timely reap-portionments of the legislature, the amendment simply expanded the content of a single, discrete component of the original constitution without altering its essential character or function. Furthermore, there is no reason to conclude that, in light of the structure of the constitution, the provisions of the 1952 amendment could not sensibly have been placed in *560a single section of the constitution. In other words, the substantive changes made by the amendment were related closely enough to be considered to be a single amendment to a single section of the constitution.

However one analyzes the decision in Baum, I have no doubt that it reached the right result under Article XVII, section 1. An interpretation of Article XVTI, section 1, that required a separate vote on any of the components of the 1952 amendment would be an interpretation that made little or no sense. That is why I am now persuaded that the test that we established in Dale v. Keisling, 167 Or App 394, 999 P2d 1229 (2000), to implement Article XVII, section 1, was wrong. We held in Dale that multiple substantive changes produced by the enactment of a measure are “closely related” if a vote in favor of one necessarily implies a vote in favor of the others. Id. at 401. Under that test, many of the provisions in the 1952 reapportionment amendment would have had to have been voted on separately because a vote for one of them would not logically imply a vote for the others. We erred in Dale because we focused on the relationship of one substantive change to another without considering the relationship of the various changes to the whole, and because we failed to recognize that the constitution’s structure bears on whether a group of substantive changes to the constitution are related closely enough to constitute a single amendment.

The Supreme Court has since confirmed in Armatta, 327 Or at 268-69, and in Hartung v. Bradbury, 332 Or 570, 579 & n 5, 33 P3d 972 (2001), that Baum was correctly decided and would not have been decided differently under the separate-vote analysis that the court established in Armatta and has applied in subsequent cases. Consequently, any application of the principles established in Armatta must be consistent with Baum.7

The decision in Hartung also is instructive. There, the court upheld a 1986 amendment to the reapportionment section that was at issue in Baum. The 1986 amendment *561essentially codified the court’s holding in McCall v. Legislative Assembly, 291 Or 663, 674-86, 634 P2d 223 (1981), that state senators whose terms continue beyond the first election that is held under a new apportionment plan must be assigned to represent districts that did not elect senators at that election. Compare id. with Or Const, Art IV, § 6(1) (1986). It also amended the constitution to provide that the senators who are assigned to the new districts are subject to recall by the electors of the new districts rather than by the electors of the former districts from which they were elected. Compare Or Const, Art IV, § 6(5) (1986), with Or Const, Art IV, § 6 (1952). The latter change is a substantive change that represents a policy choice that is closely correlated to the policy choice to assign holdover senators to new senate districts. That change also happens to involve a subject—the recall of elected public officials—that is addressed by another section of the constitution, Article II, section 18.

It makes practical and political sense to design the system so that senators assigned to new districts are subject to recall by the electors of the new districts, but the system presumably could function if holdover senators were subject to recall by the electors of their former districts rather than their new districts. See McCall, 291 Or at 676-82. In other words, the policy choice that the change made is not one that was compelled by the decision to codify the McCall holding by assigning holdover senators to new districts. Furthermore, as a matter of constitutional design or structure, the recall provision for holdover senators could have been made part of Article IV, section 6, as it was, or it could have been made part of Article II, section 18, the recall section of the constitution. A proper understanding of the close-relationship test of Armatta must consider the import of those features of the 1986 measure, because Hartung held that the substantive changes made by the 1986 reapportionment amendment were related closely enough to meet the standard imposed by the separate-vote requirement.

Several other separate-vote cases must be considered as well. Armatta is the most prominent among them. As noted earlier, Armatta held that Ballot Measure 40 (1996), violated the separate-vote requirement. Among other things, the measure made substantive changes to a number of the *562provisions of the Oregon Bill of Rights. The court concluded that changes to existing provisions of the Bill of Rights that serve very different purposes and affect different groups of people were not changes that were closely related. Armatta, 327 Or at 283-84. That makes sense. The Bill of Rights embodies a number of distinct policy choices that are the products of very different historical developments. Except for provisions of the Bill of Rights that are related to each other, such as the provisions on religious liberty, see Or Const, Art I, §§ 2-7, it is unlikely that changes to two or more provisions of the Bill of Rights could be considered to be closely related changes because, as the court later put the point in Lehman, “it is difficult to make related changes to unrelated constitutional provisions.” Lehman, 333 Or at 246. The court subsequently applied that principle in League of Oregon Cities v. State of Oregon, 334 Or 645, 673-74, 56 P3d 892 (2002), in which it held that an amendment that made substantive changes to the guarantee of free expression in Article I, section 8, and the guarantee of just compensation for the taking of property in Article I, section 18, violated the separate-vote requirement.

Lehman involved a different problem. It concerned a challenge to Ballot Measure 3 (1992), which amended the Oregon Constitution to establish or modify term limits for a number of elected state officials and for the members of Oregon’s congressional delegation. With regard to the state officials, the court rejected the plaintiffs’ contention that changes that affected legislative and executive officials were unrelated because they involved officials in different branches of government who performed very different functions. The court held that those changes were related because the original constitution had several provisions that imposed limits on elected officials in all three branches of state government. See Lehman, 333 Or at 247-48. That suggested, as a structural matter, that changes to the restrictions affecting elected officials in different branches of state government could appropriately be considered to be related to each other.

However, the court did not need to resolve whether those changes were closely related for purposes of the separate-vote requirement. That is because it concluded that the changes that affected the state officials were not closely related to those that affected the federal officials, which *563meant that the measure violated the separate-vote requirement and, hence, was invalid. Id. at 248-50. In reaching that conclusion, the court noted that, in contrast with the provisions in the original constitution that established various qualifications for elected state officials, there were no provisions in that constitution that imposed qualifications or limits on federal officials elected from Oregon. Instead, the qualifications and limits that apply to people elected to Congress from Oregon are found in the federal constitution.

There is a reason for that. The state and federal governments are fundamentally different entities. With few exceptions, decisions on how to constitute and limit state government and its elected officials are decisions that the people of Oregon make. In contrast, decisions about how to constitute and limit the federal government and its elected officials are decisions that the entire country makes. Consistent with that point, a decision to impose a limit on elected state officials presents a very different structural issue from a decision to impose a limit on elected federal officials. For example, a decision to impose term limits on all elected officials in the legislative and executive branches of state government affects both branches of state government in a consistent way. In contrast, a decision to impose term limits on the members of Oregon’s congressional delegation affects only Oregon’s members. In a Congress in which no other state imposes similar limits on its congressional delegation and in which seniority matters, the decision places Oregon in a very different position from all other states. Of course, the United States Supreme Court held in U.S. Term Limits, Inc. v. Thornton, 514 US 779, 115 S Ct 1842, 131 L Ed 2d 881 (1995), that states cannot impose term limits on the members of Congress because they lack the authority to add to the qualifications in the federal constitution for members of Congress. For our purposes, the important point is that a change in the Oregon Constitution on the qualifications for elected state officials presents an issue that is qualitatively and structurally different from a change in the qualifications for Oregon’s congressional delegation. Lehman properly recognized that fundamental difference.

The majority appears not to understand my point about Lehman. See 188 Or App at 550-52. As noted, U.S. *564Term Limits established that a provision imposing term limits on members of Congress cannot properly be made a part of a state constitution, which serves to emphasize the point that I have tried to make—that state policies affecting state officials present very different issues from those affecting federal officials. The state and federal officials are chosen by the same electors, but they perform functions for separate sovereigns. Even if a state constitution could be amended to include a provision on the qualification of members of Congress, the amendment that made that change would have to be adopted separately from one addressed to the qualifications of state elected officials because, as a matter of constitutional structure, state and federal officials are fundamentally distinct. Although the policy goal and subject of the two amendments would be the same—limiting the term óf office of people elected to public office—the state officials affected by the policy would be distinct from the federal officials affected by the policy when viewed in terms of the structure of the state and federal governments and, consequently, the structure of the relevant constitutions.

The final case that bears on the issue is Swett v. Bradbury, 333 Or 597, 43 P3d 1094 (2002). It involved a challenge to Ballot Measure 62 (1998), which adopted a number of provisions on the conduct of Oregon elections. The court focused on two of those provisions in concluding that the measure violated the separate-vote requirement. One of the provisions implicitly amended Article II of the constitution by imposing new campaign finance disclosure requirements on elected public officials who receive more than $500 from any one contributor. The other provision amended Article IV by adding a requirement that people who gather initiative signatures be registered Oregon voters. The court had little difficulty concluding that those two provisions were not closely related. The two provisions involved very different subjects— the conduct of elected public officials versus the qualifications of people who solicit initiative signatures—that are found in two different articles of the original constitution and that bear no apparent relationship to each other in terms of constitutional structure or design. See Swett, 333 Or at 608-09.

I will now turn to the measure involved in this case, Ballot Measure 3 (2000). It adopts a number of provisions *565that deal with the forfeiture of property to the government. The use of civil forfeiture as a law enforcement tool is a relatively recent phenomenon. Although a body of law has developed over the past several decades concerning its use, it remains an innovation whose use has expanded greatly over the past 30 years. Measure 3 represents an effort to modify and systematize the use of civil forfeiture in Oregon. As such, it contains a number of provisions that make substantive changes to the constitution in order to establish a systematic policy on civil forfeiture. They include (1) a provision that requires that property owners be convicted of crimes and their property be established "to have been instrumental in committing or facilitating the crime or to be proceeds of that crime” before their property can be forfeited to the government, (2) a provision that seeks to protect financial institutions and others who have interests in property that is subject to forfeiture, (3) a provision that addresses the forfeiture of unclaimed property and contraband, (4) a provision that preserves the ability of the state to retain property temporarily for evidentiary or protective purposes, (5) a provision that governs the sale of forfeited property and the distribution of the funds obtained from forfeiture, (6) provisions that address the manner in which the state deals with the federal government on forfeiture, and (7) a provision that imposes civil penalties against people who violate specified provisions of the measure. See Or Const, Art XV, § 10.8

Because the measure makes several substantive changes to the constitution, the issue for us is whether those changes are closely related enough for purposes of the separate-vote requirement to be adopted in a single amendment. I am persuaded that they are.

In my view, Measure 3 is equivalent to the 1952 reapportionment measure that was upheld against a separate-vote challenge in Baum. Both measures adopted a group of provisions that were intended to establish a coherent and comprehensive scheme to address a discrete policy goal.9 The *5661952 reapportionment measure sought to establish a system by which to ensure periodic reapportionment of the Oregon legislature. The measure embodied a series of policy decisions about the manner in which to accomplish that goal, but those policy decisions were ones that had to be addressed in order to accomplish the goal. There was no reason in terms of the structure of state government or the structure of the constitution to require the various provisions of that measure to be adopted by separate votes.

The same principles apply to Measure 3. It seeks to establish a comprehensive approach to the use of civil forfeiture as a law enforcement tool.10 It embodies a series of policy decisions about civil forfeiture, but those policy decisions are ones that had to be addressed in order to establish a comprehensive system of civil forfeiture. Furthermore, they are not decisions that, in terms of the structure of state government or the structure of the constitution, needed to be submitted to separate votes.11

Admittedly, the measures struck down in Armatta, League of Oregon Cities, Lehman, and Swett also sought to establish comprehensive systems of some kind, so it is not enough to show that the provisions of an amendment serve such a purpose. That showing may satisfy the single-subject *567requirement of Article IV, section 1(2)(d), but the separate-vote requirement addresses a different concern.

The nature of the latter concern is perhaps best addressed here through consideration of the two changes to the constitution on which the majority focuses in concluding that Measure 3 violated the separate-vote requirement. Those changes are (1) the addition of a requirement that a property owner be convicted of a crime in order for the government to forfeit the owner’s property and (2) the addition of a requirement that funds obtained by forfeiture be used to fund drug treatment programs. 188 Or App at 547. Although those two policy choices could have been submitted to the voters separately, there is no reason to believe that Article XVII, section 1, was intended to require them to be submitted to the voters separately. In particular, there is no reason in terms of the structure of state government or of the constitution to conclude that a provision that creates or regulates a source of funds must be adopted separately from a provision that controls the use of those funds. The 1984 measure that created the state lottery, Ballot Measure 4 (1984), contained provisions that created and regulated the state lottery and specified how the money raised from the lottery was to be spent by the state. Although the combination of those changes to the constitution in a single amendment has not been subjected to a separate-vote challenge, I believe that such a challenge, if made, would not succeed. I believe that the same conclusion applies to the two provisions on which the majority focuses in this case.

In reaching a different conclusion, the majority says that the changes at issue here appear to be no more closely related than were the changes at issue in Lehman, which the Supreme Court held to violate the separate-vote requirement. The majority says:

“If the constitutional changes at issue in Lehman, both of which related to a specific objective—the creation of term limits for public officials—had ‘little or nothing to do’ with each other, Lehman, 333 Or at 250, then it is difficult to conceive how the constitutional changes effected by subsections 3 and 7 of Measure 3 could be regarded as closely related. There is little, if any, logical relationship between subsection 3, a provision that gives a property owner the *568right to be convicted of a crime before his or her property may be forfeited, and subsection 7, which directs and restricts the purposes for which forfeiture proceeds can lawfully be used.”

188 Or App at 547. The comparison is inapt.

As noted earlier, the qualifications of elected state officials present very different issues in terms of the structure of state government and of the state constitution from those of the members of Oregon’s congressional delegation. Although the term-limits measure sought to achieve a specific objective, the relevant provisions of that measure addressed the qualifications of two different groups of elected officials that, in structural and constitutional terms, bore no relationship to each other. The changes on which the majority focuses in this case are in no sense comparable.

The mistake that the majority makes is to confuse the separate-amendment and single-subject requirements of the constitution. As I have explained, the separate-vote requirement is concerned with the structure of state government and the structure of the constitution. Whether two changes are closely related for purposes of that requirement concerns how the changes fit together when viewed in terms of governmental and constitutional structure. The focus is not on whether the changes achieve policy goals that are closely related to each other.

Contrary to the majority’s suggestion, see 188 Or App at 549-50, I recognize that an amendment that runs afoul of the single-subject requirement will necessarily run afoul of the separate-vote requirement because an amendment cannot satisfy that latter requirement if it includes provisions addressed to two different subjects. What the majority fails to understand, however, is that the question whether two constitutional changes are closely related for purposes of the separate-vote requirement does not concern whether the two changes are closely related as subjects, but whether they are closely related in a way that makes it appropriate for them to be included in a single, discrete provision of the constitution, in light of how the constitution, itself, is structured. *569The focus of the single-subject and separate-vote requirements is different in a way that the majority simply does not appreciate.

A comparison of the reapportionment amendment upheld in Baum and the forfeiture amendment at issue here should help clarify the point. In order to ensure that the legislature is reapportioned every 10 years, there must be a mechanism created to ensure that that occurs. There also must be a formula by which to allocate legislators to the different parts of the state. The 1952 reapportionment amendment addressed both issues. It provided for direct Supreme Court review of reapportionment plans, it required the Secretary of State to adopt a reapportionment plan if the legislature failed to adopt a valid plan, and it adopted a formula to be used to allocate legislators to different parts of the state. See Or Const, Art IV, § 6 (1952).

When viewed in isolation, there is no logical connection between the constitutional change that imposed on the Secretary of State the duty to adopt a valid reapportionment plan if the legislature failed to do so and the constitutional change that adopted a reapportionment formula for the allocation of legislators in the state.12 Nevertheless, Baum held that those changes were properly adopted as part of a single amendment, which means that the changes were closely related for purposes of the separate-vote requirement. That conclusion has to be correct. As a matter of constitutional design or structure, a reapportionment provision in a constitution is the type of provision that would commonly be placed in a single section of the constitution. Components that properly comprise such a provision, such as those addressed to the mechanisms and formula used to reapportion the legislature, are components that are closely related to each other for purposes of the separate-vote requirement.

The same analysis applies to the forfeiture amendment. A forfeiture system must contain provisions that specify the property and interests in property that are subject to *570forfeiture and the use to be made of forfeiture proceeds. Measure 3 addressed both issues. It identified the property that is subject to forfeiture and provided that property owners must be convicted of a crime in order to have their property forfeited. It also required that the funds obtained from forfeiture be used for drug treatment.

When viewed in isolation, there is no logical connection between the constitutional change that requires that property owners be convicted of a crime in order for their property to be subject to forfeiture and the constitutional change that requires that forfeiture proceeds be used for drug treatment.13 Nevertheless, as in Baum, those changes are closely related for purposes of the separate-vote requirement. As a matter of constitutional design or structure, a provision addressed to the forfeiture of property by the government is the type of provision that would commonly be placed in a single section of the constitution. Components that properly comprise such a provision, such as those addressed to the conditions under which forfeiture can occur and the use to be made of forfeiture proceeds, are components that are closely related to each other for purposes of the separate-vote requirement.

The majority also suggests that the two changes are not closely related for purposes of the separate-vote requirement because they arguably confer or address rights affecting different groups of people:

“In determining whether the provisions are closely related, we also consider whether they affect the rights of *571different groups of people. Here, they arguably do. Subsection 3 confers a right on owners of property subject to possible forfeiture, for whom a criminal conviction must precede forfeiture. Subsection 7, on the other hand, arguably confers a benefit on a different group, persons receiving drug treatment.”

188 Or App at 547-48. The majority is wrong. The provision that restricts the use of forfeiture proceeds to programs that provide drug treatment does not confer on people who need drug treatment a right to receive drug treatment any more than the creation of a dedicated fund for any other activity is understood to create rights in the possible beneficiaries of the fund. In other words, if the majority were correct, then the creation of the highway trust fund or the original lottery fund created rights in favor of the potential beneficiaries of those funds. Whatever ability potential beneficiaries (or taxpayers for that matter) have to obtain court relief against a misapplication of those funds, they do not have rights that are equivalent to the rights that are conferred on people by the Bill of Rights, which are the type of rights that were at issue in Armatta that led the court to invalidate Measure 40 on separate-vote grounds.14 It is the latter type of rights that, as a matter of constitutional structure, are unrelated to one another and that, as a consequence, must be addressed through separate amendments.

In summary, although Measure 3 made a number of substantive changes to the Oregon Constitution, those changes are closely related changes when evaluated under the policies that Article XVII, section 1, seeks to achieve. The majority errs in concluding otherwise. In light of the nature of Measure 3,1 also have no difficulty concluding that *572it complied with the single-subject requirement in Article IV, section l(2)(d), of the constitution. I therefore would affirm the trial court’s judgment that declared that Measure 3 complied with the separate-vote and single-subject requirements of the constitution.