specially concurring.
I concur with the decision of the plurality that the Court of Appeals erred in holding that Ballot Measure 3 (2000) was invalid under Article XVII, section 1, of the Oregon Constitution, which provides, in part:
“When two or more amendments shall be submitted in the manner aforesaid to the voters of this state at the same *514election, they shall be so submitted that each amendment shall be voted on separately.”
However, I do not agree with all the reasoning that the plurality employs to reach its conclusion. I write separately to explain my reasons for agreeing with the plurality’s ultimate conclusion.
The plurality notes that the question here, i.e., whether Measure 3 contains two or more amendments to the Oregon Constitution, is a “recurring issue.” 341 Or at 504.1 agree that Oregon’s courts and many litigants continue to struggle with the correct application of Article XVII, section 1. The problem, however, lies at least in part in the difficulty that judges and litigants face in applying this court’s case law under that provision. Inconsistent and unexplained conclusions and analytical models based on subjective criteria exacerbate that problem. I discuss below some of those sources of confusion. Although analytical simplicity in this area probably is not possible, the court must be willing to examine its evolving case law to ensure that the various “tests” that it has developed continue to apply the constitution’s terms faithfully. It is in the interest of that ultimate goal — accurate constitutional interpretation — that I offer the observations below.
This court has noted that Article XVII, section 1, partially shares the objective of another constitutional provision, Article IV, section l(2)(d), which provides, in part:
“A proposed law or amendment to the Constitution shall embrace one subject only and matters properly connected therewith.”
In Armatta v. Kitzhaber, 327 Or 250, 275-76, 959 P2d 49 (1998), this court stated with regard to the separate-vote and single-subject requirements:
“First, the purposes behind the two requirements are similar: Both serve to ensure that the voters will not be compelled to vote upon multiple ‘subjects’ or multiple constitutional changes in a single vote.
“However, it is significant that, from the beginning of statehood, the single-subject and separate-vote requirements have been worded differently. As we have discussed, *515the single-subject requirement * * * focuses upon the content of a proposed law or amendment, by requiring that it embrace only one subject and matters properly connected therewith. * * *
“The separate-vote requirement, by contrast, focuses upon the form of submission of an amendment, as well as the potential change to the existing constitution, by requiring that two or more constitutional amendments be voted upon separately. That is, in addition to speaking to the form of submission, the separate-vote requirement addresses the extent to which a proposed amendment would modify the existing constitution. That is significantly different from the wording of the single-subject requirement, which focuses in isolation upon only the text of a proposed amendment in requiring that it embrace a single subject.
“* * * Indeed, because the separate-vote requirement is concerned only with a change to the fundamental law, the notion that the people should be able to vote separately upon each separate amendment should come as no surprise. In short, the requirement serves as a safeguard that is fundamental to the concept of a constitution.”
(Emphases in original.)
Consistently with that passage from Armatta, we must bear in mind the different ways in which the separate-vote and single-subject provisions apply. In particular, we must endeavor not to blur the distinctive protections that those provisions impose on the process of amending the constitution by initiative.
Armatta drew attention to the fact that the constitution does not define the term “amendment”:
“Although Article XVII, section 1, does not define what is meant by ‘two or more amendments/ it is important to note that the text focuses upon the potential change to the existing constitution, by requiring that two or more constitutional amendments be voted upon separately.”
Id. at 263 (emphases in original). That statement is undoubtedly correct. However, it leaves open the question of the correct application of Article XVII, section 1, in two separate contexts that are pertinent to the problem in this case: (1) an amendment that modifies or repeals existing constitutional *516wording either expressly or by implication; and (2) an amendment that adds new wording to the constitution that does not modify or otherwise affect the operation of existing constitutional provisions.
The Armatta court’s discussion of two early Oregon cases sheds at least some light on the answer to that issue. The court noted that State of Oregon v. Payne, 195 Or 624, 635, 244 P2d 1025 (1952), confirmed that “the fact that a proposed constitutional amendment contains more than one section does not preclude its submission as a single amendment.” Armatta at 268.
Additionally, in. Armatta the court observed that, in Baum v. Newbry et al., 200 Or 576, 581, 276 P2d 220 (1954), the court had stated:
“[The separate vote requirement] does not prohibit the people from adopting an amendment which would affect more than one article or section by implication. * * * 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.”
In summarizing the Baum holding, the court in Armatta stated:
“Baum stands for the following principles. 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 *517extent, the notion that a single amendment must contain a single ‘subject.’ ”
Armatta, 327 Or at 269.
The conclusion that Armatta drew from Payne, quoted above, seems unremarkable: A permissible single amendment may contain multiple sections. The summary of Baum in Armatta is more problematic, in part because it repeated certain conclusory, ambiguous statements of the court in Baum. For example, both Baum and Armatta indicate that a single constitutional amendment may “affect” one or more constitutional provisions without violating the separate-vote requirement. Baum, 200 Or at 581; Armatta, 327 Or at 269. That does not explain, however, whether the term “affect” refers to (1) a modification or repeal of one or more existing constitutional provisions; or (2) an addition of wording to an existing constitutional provision without repealing or modifying existing wording; or (3) both of the above.
Moreover, it appears that the Armatta court, in summarizing the final principle that it drew from Baum, i.e., “a single amendment must contain a single ‘subject!,]’ ” 327 Or at 269, modified the statement in Baum that “[a]t most [the separate vote requirement] 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. That restatement of the holding in Baum is notable, and correct in my view, because it did not repeat the notion from Baum that the separate-vote requirement applies solely to multiple amendments that concern different “subjects.”1 The constitutional source for the requirement that a proposed amendment contain one subject and *518matters properly connected therewith is Article IV, section l(2)(d), not Article XVII, section 1.
Armatta itself illustrates that point. In Armatta, the proposed constitutional amendments changed the effect of the existing terms of several constitutional provisions. The court held that the proposal violated the separate-vote requirement. Armatta, 327 Or at 284. The fact that the proposed constitutional amendments probably were germane to one subject, i.e., the rights of crime victims, and thus probably satisfied the single-vote requirement in Article IV, section l(2)(d), was beside the point. As the court in. Armatta put it,
“Although the court in Baum referred to a hypothetical amendment containing multiple ‘subjects,’ the court did not state that, if a proposed amendment contains a single subject, then it also must be deemed to be a single amendment.”
Id. at 274. Armatta went on to expressly reject the state’s argument that a proposed amendment satisfies the separate-vote requirement if it satisfies the single-subject requirement. Id. at 277.
After observing that the prior Oregon cases were “lacking in detailed analysis[,]” id. at 275, the Armatta court stated:
“We conclude that the 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 substantive 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.”
Id. at 277. The court had no difficulty applying that test in Armatta, because the initiative measure clearly changed the substance of numerous provisions of the state constitution.
Because Armatta was an “easy” case, the court spent no time in its opinion attempting to explain how it arrived at the two announced separate-vote criteria “substantive” and *519“closely related,” or how they applied. However, it soon became evident that those criteria were causing confusion.
In Dale v. Keisling, 167 Or App 394, 999 P2d 1229 (2000), the Court of Appeals opined that, to satisfy the two criteria announced in Armatta, two or more substantive changes to the constitution would have to be so closely related that a vote in favor of one proposed amendment necessarily would imply a vote in favor of the other. Id. at 401.1 refer to that test as the “necessary implication” test.
In 2002, this court addressed a challenge to Measure 3 (1992) under the separate-vote provision. In Lehman v. Bradbury, 333 Or 231, 37 P3d 989 (2002), this court concluded that Measure 3, which concerned term limits for various public officials, embodied multiple amendments to the constitution in violation of the separate-vote requirement.
The trial court in Lehman had attempted to apply the “necessary implication” test from Dale. On appeal, the Secretary of State complained that the “closely related” criterion from Armatta required clarification. Id. at 242. The Secretary of State urged the court to adopt the following test for determining whether multiple amendments to the constitution are “closely related”:
“[T]wo or more changes to the constitution are ‘closely related’ if they are so logically interrelated as to present one specific, discrete, cohesive policy choice.”
Id.
This court refused to accept the test that the Secretary of State offered, stating:
“Defendant apparently believes that Armatta needs clarification. However, adopting defendant’s ‘clarification’ would mean that we potentially were permitting our task under Article XVII, section 1, to degenerate into an endless war of adjectives and adverbs, each battle of which would involve further efforts to explain and elaborate on whichever set of adjectives and adverbs has been used in the next preceding case. That does not mean that we would not accept a party’s proposed reformulation of an existing analytical test, if it appeared that the proposed test would be a *520better tool to use in future cases. Defendant’s proffered test simply does not appear to us to be a better tool.”
Id. Lehman rejected the “necessary implication” test from Dale as well as the adjective-laden test that the Secretary of State offered.
I agree that those proffered tests were flawed. In addition to their subjective and, thus, standardless character, they did not reflect the core requirement that Armatta correctly had discerned in the separate-vote requirement: A measure must embody “a particular constitutional change” that will allow voters to “express their will in one vote as to only one constitutional change.” Armatta, 327 Or at 269.
Lehman easily concluded that Measure 3 proposed multiple substantive changes to the wording of different provisions of the constitution. Lehman, 333 Or at 244. The analysis could have stopped there, because, under the core requirement of the separate-vote rule, as this court construed that rule in Armatta, Measure 3 failed to offer voters “a particular constitutional change” about which the voters could express their will in one vote.
However, Lehman proceeded to inquire whether the multiple substantive constitutional amendments were “closely related.” Id. The court acknowledged that Armatta had not explained what the “closely related” criterion meant. Id. However, the court took note of several of the observations that the Armatta court made about the proposed measure that it examined and concluded that those comments were themselves additional legal tests regarding the “closely related” criterion.
As a consequence of that conclusion, Lehman announced that the court would apply the “closely related” criterion on two separate levels. First, the court would ask whether the multiple constitutional provisions that the measure modified are themselves “closely related.” Id. at 246. Second, the court would ask whether the constitutional changes embodied in the proposed measure are themselves “closely related.” Id. After applying those multiple tests, the court concluded that Measure 3 violated the separate-vote requirement. Id. at 250.
*521The court’s ultimate conclusion in Lehman remains persuasive to me. In many ways, Measure 3 was as plainly violative of the separate-vote requirement as was the measure addressed in Armatta. However, the Lehman court’s elaboration of the “closely related” criterion deserves reconsideration for several reasons.
First, Article XVII, section 1, prohibits the submission to the voters of more than one amendment to the constitution for a single vote. It does not invite the submission of multiple amendments for one vote if judges decide that the affected constitutional provisions or the proposed amendments themselves have a relationship that is “close.” By opening the door to the submission of multiple substantive constitutional amendments for a single vote, the court risks the emasculation of the important protection that Article XVII, section 1, embodies.
Second, the court’s reliance on a test that incorporates multiple applications of a subjective, court-created phrase simply feeds an unfortunate public perception that judges execute only their personal predilections in applying the constitution. One can hardly imagine a phrase more elastic, or more lacking in some objective foundation, than “closely related.” The court originally (and correctly) intended that criterion as a means of disposing of the argument that any change to more than a single word in the constitution would violate Article XVII, section 1. As Lehman stated,
“[I]n any separate-vote inquiry, it is imperative that we remain aware that any amendment to the constitution involves some change to the wording of that document. However, not every one-word change to the wording of the constitution is a separate ‘amendment.’ If it were, then amendments to the constitution would have to happen word-by-word, and the people’s power to amend the constitution would be hamstrung.”
Id. at 240. The court should restore that focus to its analysis of whether a proposed change in constitutional wording is a separate amendment.
Third, asking whether several proposed amendments to different constitutional provisions share a “close *522relationship” is, in substance, an inquiry into whether the proposals embrace one subject and matters properly connected to that subject. Stated differently, any attempt to compare and contrast the content of multiple proposed amendments to multiple constitutional provisions will quickly devolve into an effort to discover a common theme or policy choice that the component parts tend to promote. The differences in the points of discussion offered by the plurality and the dissent in this case are a good example. The respective opinions differ over whether the elements of the instant measure are or are not germane to a common subject. But that inquiry properly results from the single-subject provision, not the separate-vote provision. The plurality’s present conception of the “closely related” test thus blurs the line that separates those distinctive constitutional requirements. To repeat, we must avoid confusing those separate legal requirements.
Two later cases further illustrate the difficulty that litigants face in attempting to comply with the court’s “closely related” criterion. In Swett v. Bradbury, 333 Or 597, 43 P3d 1094 (2002), the defendants sought to defend a campaign finance disclosure measure from a challenge under Article XVII, section 1. They agreed that the proposed measure changed the operation of multiple provisions of the constitution. Id. at 607. However, they endeavored to point out that the measure’s provisions shared a common subject, i.e., “ ‘they are regulations designed to prevent, control, or expose the influence of money in the initiative [and] referendum * * * process.’ ” Id. at 608. The court began by noting, correctly, that:
“More importantly, however, defendants’ argument fails because it is an attempt to show that sections 1 and 3 of Measure 62 share the same subject matter. 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 at 241-42 (separate-vote requirement, in contrast to single-subject requirement, focuses on extent to which proposed amendment modifies existing constitution).”
*523Id. at 609 (emphasis in original). The court also noted, correctly, that, as the parties agreed, the proposed measure altered the wording of more than one provision of the constitution. But the court then went on to assess whether those multiple alterations of the constitution bore a “close” relationship to each other. Despite the court’s attempted explanation to the contrary, that assessment consisted of determining only whether the disparate changes nevertheless shared some similar policy goal or subject. The court concluded that no such relationship was present. To adopt the court’s own words, that inquiry was “beside the point.” Id.
In Meyer v. Bradbury, 341 Or 288, 142 P3d 1031 (2006), this court decided a separate-vote question involving a proposed measure that changed the substantive operation of two different constitutional provisions. However, the court majority concluded that the measure’s alteration of the power of the Oregon legislature to enact laws regulating campaign finance was “closely related” to the measure’s alteration of the people’s constitutional right of free speech. Id. at 301. Even though the measure would not permit the voters to vote separately on those multiple, important changes to the constitution, the majority’s discovery of a “relationship” that was “close” allowed it to sidestep the separate-vote requirement.
The plurality and the dissent in this case continue that kind of debate. They differ about whether the various elements of Measure 3 do or do not share a policy theme or advance a discernible lawmaking objective. But any constitutional amendment of any degree of complexity (and most legislation for that matter) almost always will embody multiple policy objectives. Analyzing a measure for the common objectives among its parts only amounts to an assessment of whether it embraces one subject and properly connected matters, not whether it contains more than one particular constitutional change, as Armatta stated the issue.
The plurality is correct in observing that Measure 3 adds new wording to the constitution without altering the operation of any existing provision of the constitution. The dissent’s claim that Measure 3’s parts affect several constitutional provisions, or analogues of those provisions, is *524unpersuasive. Measure 3 creates one new, albeit complex, amendment to the existing constitution. The fact that Measure 3 contains several clauses that address distinctive policy goals that are germain to its subject is beside the point.
The plurality continues to search for close relationships between the parts of Measure 3. That search fails to advance the separate-vote inquiry. The fact that Measure 3’s parts contain “a unifying principle logically connecting all provisions” in the amendment demonstrates only that Measure 3 satisfies the single-subject requirement. See State ex rel Caleb v. Beesley, 326 Or 83, 91, 949 P2d 724 (1997) (stating test for single-subject requirement). Rather, Measure 3 satisfies the separate-vote requirement because it embodies one constitutional change even though that change appears in a proposal with multiple parts. Measure 3 does not compel voters to engage in a single vote on two or more changes to the existing state constitution. It does not combine in one measure multiple alterations of the constitution’s terms, either expressly or by implication. Thus, Measure 3 satisfies Article XVII, section 1.
I also join in the plurality’s conclusion that Measure 3 passes muster under the single-subject requirement in Article IV, section l(2)(d).
For the reasons stated above, I concur in the plurality’s decision to reverse the Court of Appeals and to affirm the judgment of the trial court.
Notwithstanding the selective restatement in Armatta of the comments in Baum, this court said in Hartung v. Bradbury, 332 Or 570, 579 n 5, 33 P3d 972 (2001), that it declined the petitioners’
“invitation to revisit this court’s decision in Baum in light of Armatta v. Kitzkaber * * *. Contrary to petitioners’ arguments, nothing in Armatta suggests that Baum was decided incorrectly; indeed, Armatta cites Baum favorably for the proposition that 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.’ Armatta, 327 Or at 269.”
It is true that Armatta neither overtly criticizes nor overrules Baum. However, the court’s summaries of the Baum case in. Armatta and Hartung demonstrate the court’s disinclination to endorse all that Baum had to say about the scope of the separate-vote requirement.