Chief Justice MULLARKEY
delivered the Opinion of the Court.
Pursuant to section 1-40-119, 1 C.R.S. (1998), the Secretary of State Victoria Buckley (Secretary) brought this direct appeal from the judgment of the Denver District Court which ordered her to certify initiative 1997-98 # 40, “Medical Use of Marijuana,” for the 1998 general election. In so ordering, the district court rejected the Secretary’s request that she be permitted to conduct a line-by-line determination of the sufficiency of the signatures submitted with the petition. The court equated errors made by the Secretary when issuing a statement pursuant to section 1-40-118(1), 1 C.R.S. (1998) with the secretary of state’s “fail[urej to issue a statement within thirty calendar days.” § 1-40-118(1). Thus, the court invoked the statutory remedy for inaction by the Secretary and ordered the initiative to be placed on the ballot. See id.
Because the election date was fast approaching, the ballots already had been printed, and such ballots included this initiative, we issued an order summarily reversing the judgment of the district court. We directed that the Secretary conduct a line-byline review of the signatures and that the votes cast for the initiative be counted only if the signatures were found sufficient. Our order stated that this opinion would follow.
I. Background
The facts underlying this case are not in dispute. Respondent Martin H. Chilcutt, a registered elector, is a proponent of a citizen-initiative petition that would amend the Colorado Constitution to authorize specified medical uses of marijuana for persons suffering from certain medical conditions. On July 7, 1998, he and other proponents timely submitted their initiative petition, designated 1997-98 # 40, to the Secretary for verification and determination of sufficiency of the petition signatures in order to certify the initiative for the 1998 general election. The constitutional formula required 54,242 signatures for an initiative to qualify for the 1998 general election ballot. The Secretary conducted a random sampling of 4,482 of the 88,815 signatures submitted. Her projection, based on the random sample, indicated that the petition contained 47,960 valid signatures or 88% of the requisite number to certify the Medical Use of Marijuana measure for the general election. Because the random sample indicated that the petition contained less than 90% of the required number of signatures, on August 6, 1998, the Secretary issued a statement determining that the petition was insufficient pursuant to section 1-40-116(4), 1 C.R.S. (1998).
On September 4, 1998, Chilcutt filed a timely protest in Denver District Court as permitted by section 1-40-118, 1 C.R.S. (1998). Following the filing, the Secretary conducted a supplementary review. A hearing was held on September 11, 1998 at which Chilcutt argued that the Secretary had erred in her initial review of the signatures. Chil-cutt asserted that a correct assessment of the random sample indicated that the petition contained 96.4% of the required number of signatures. The Secretary, as a result of her supplementary review, conceded that fifty signatures previously declared invalid were in fact valid and that, based on this correction, the petition contained more than 90% of the required signatures. The Secretary acknowledged that the initial statement determining the petition to be insufficient was issued in error and stipulated to this fact before the district court.
The Secretary argued that, pursuant to section 1-40-116(4), the law required her to conduct a line-by-line count. Because the deadline for certification of questions to the ballot was September 14, 1998, the Secretary offered to certify the measure to the ballot [115]*115while she conducted the count. If the line-by-line count resulted in a conclusion that the measure contained a sufficient number of signatures, then the votes cast for the measure would be counted. If the measure did not contain the required number of signatures, then the votes would not be counted. Under this proposal, the proponents would retain the right to protest a new determination of insufficiency. According to the Secretary’s proposal, if the proponents prevailed on an appeal, then the measure would be placed on the ballot in the year 2000, the year of the next general election.
At the protest hearing, the district court concluded that there was “no express authority for the secretary of state to do a line-byline analysis outside the statutory parameters.” Chilcutt v. Buckley, No. 98CV6940, slip op. at 15 (D.Ct. Sept. 11, 1998). In addition, the district court rejected the Secretary’s proposal to print the initiative on the ballot while conducting a line-by-line count for certification. The district court reasoned that such a process would be “cumbersome, awkward, and messy” in addition to generating significant uncertainty. Id. at 16. Applying these concerns to the factors we set out in Rathke v. MacFarlane, 648 P.2d 648, 652-53 (Colo.1982), the district court concluded that injunctive relief was appropriate and ordered the Secretary to certify the initiative for the ballot because of the state’s policy “to favor enfranchising the voters of this state rather than creating potential disenfranchisement.” Chilcutt, No. 98CV6940, slip op. at 16.
II. Analysis
A. The Initiative Process
We begin our analysis with relevant background regarding the initiative process. The process is governed by Colorado Constitution article V, section 1 and sections 1-40-101 to 134, 1 C.R.S. (1998). Initiative proponents circulate copies of an initiative petition in order to accumulate the constitutionally required number of signatures from “registered electors.” See Colo. Const, art. V, § 1(2); § 1-40-111(1). In order to qualify for the ballot, an initiative must have signatures from registered electors totaling “at least five percent of the total number of votes cast for all candidates for the office of secretary of state at the previous general election.” Colo. Const, art. V, § 1(2). A petition required 54,242 signatures to qualify for the ballot in the 1998 general election. The Secretary is required to verify the signatures on the initiative petition. See § 1-40-116(2). The Secretary conducts the first step of the verification process through random sampling. See § 1-40-116(4).
If the random sample indicates that the petition as a whole contains less than 90% of the required number of valid signatures, then the statute deems the petition to be “not sufficient.” See id. If the random sample indicates that the number of valid signatures equals or exceeds 110% of the number required for certification, then the statute deems the petition “sufficient.” See id. If the random sample indicates that the number of valid signatures is greater than 90% but less than 110% of the requisite number, then the Secretary is required to “order the examination and verification of each signature filed.” Id. If this “line-by-line” count indicates that the petition contains the constitutionally required number of signatures, 54,-242 for the 1998 election, the initiative must be submitted to the people “for adoption or rejection at the polls.” See Colo. Const, art V, § 1(7). The Secretary’s authority to conduct a line-by-line examination is at issue in this case.
Article V, section 1 of the Colorado Constitution and sections 1-40-116 to -119, 1 C.R.S. (1998) govern the timing of the initiative process. Article Y, section 1(2) of the constitution requires that an initiative petition must be filed with the Secretary at least three months before the general election on which it is to be voted upon. Once a petition has been filed, the Secretary is required to issue a determination of sufficiency of the signatures within thirty days. See § 1-40-118(1). If the Secretary finds the number of signatures is insufficient, the proponents have fifteen days to cure the insufficiency provided the new signatures are submitted at least three months before the election. See § l-40-117(3)(b). The cure option was not available to the initiative proponents in this [116]*116ease because they could not meet the three-months deadline.
A statutory remedy is provided for the Secretary’s failure to act within thirty days: “If the secretary of state fails to issue a statement within thirty calendar days, the petition shall be deemed sufficient.” § 1-40-118(1). If the Secretary issues a statement of sufficiency, any registered elector may file a protest within thirty days after the Secretary issues the statement. See id.
Nothing in the statute expressly addresses the situation now before us. The Secretary admittedly made mistakes and excluded some valid signatures when she randomly sampled the signatures. However, when the errors were corrected, the sample still contained insufficient signatures to automatically qualify the initiative for the ballot under section 1-40-116(4). Rather, the corrected random sample came within the range (more than 90% but less than 110%) of section 1-40-116(4) that requires a line-by-line count of all signatures. The protest section of the statute, section 1-40-118(2), allows any registered elector to challenge defects in the Secretary’s random sample, but the section is silent as to the remedy available under the circumstances of this case.
For guidance in construing the statutes, we turn to the constitution and basic principles of statutory construction.
B. Constitutional Requirements
Our first consideration regarding the consequences of the Secretary’s sampling error and her subsequent determination that more than 90% but less than 110% of the signatures were valid is a constitutional one. On the one hand, our analysis must be guided by the principle that legislation concerning a constitutional right like the power of initiative must not impair that right:
Although constitutional provisions which are self-executing require no implementing legislation, legislation that furthers the purpose of self-executing constitutional provisions or facilitates their enforcement is permissible. However, legislation which directly or indirectly impairs, limits or destroys rights granted by self-executing constitutional provisions is not permissible.
Zaner v. City of Brighton, 917 P.2d 280, 286 (Colo.1996) (quoting Brownlow v. Wunsch, 103 Colo. 120, 123, 83 P.2d 775, 777 (1938)). On the other hand, our analysis is bounded by the principle that we must presume a statute to be constitutional, see Harris v. Heckers, 185 Colo. 39, 41, 521 P.2d 766, 768 (1974); accord In re Marriage of Franks, 189 Colo. 499, 506, 542 P.2d 845, 850 (1975), and that if two interpretations are possible, we must avoid an interpretation that might render a statute unconstitutional. See Meyer v. Putnam, 186 Colo. 132, 134, 526 P.2d 139, 140 (1974).
The purpose for a detailed signature verification procedure is “to maintain integrity in the initiative process and to comply with the constitutional requirements.” Fabec v. Beck, 922 P.2d 330, 335 (Colo.1996) (citing Colo. Const, art. V, § 1(6)) (emphasis added). The process should “properly safeguard, protect, and preserve inviolate for [the people] these modern instrumentalities of democratic government.” § 1-40-101, 1 C.R.S. (1998).
A petition meeting constitutional requirements that has been properly verified “shall be prima facie evidence that the signatures thereon are genuine and true and that the persons signing the same are registered electors.” Colo. Const, art V, § 1(6); see also § 1-40-116(1). If the proponents have gathered the constitutionally requisite number of signatures, the Secretary must submit the proposed measure to the people “for adoption or rejection at the polls.” Colo. Const, art. V, § 1(7).
Chilcutt asserts that the Secretary does not have any implied authority to act in this area unless that authority would “facilitat[e] rather than frustrate] the exercise of the right of initiative.” He suggests that our previous decisions in Fabec and Montero v. Meyer, 795 P.2d 242, 246-47 (Colo.1990), support this conclusion because both emphasized liberal construction regarding compliance with the initiative process.
While Fabec and Montero emphasize liberal construction, they do so primarily with respect to technical statutory considerations: “The constitutional right reserved to the peo-[117]*117pie may be facilitated, and not hampered by either technical statutory provisions or technical construction thereof.” Montero, 795 P.2d at 245 (emphasis added). In Montero, we held that resubmission of petitions that had been technically “withdrawn” in an effort to cure a signature insufficiency was not subject to the initial petition deadline specified in the Colorado Constitution. See Montero, 795 P.2d at 246. The constitution requires petitions to be filed with the Secretary “at least three months before, the general election at which they are to be voted upon.” Colo. Const. art. V, § 1(2). Our decision permitted the initiative proponents the time statutorily provided under the then existing cure provisions to resubmit their petition. See Montero, 795 P.2d at 245-46.1
At issue here is not a technical procedural provision but a substantive constitutional requirement. A ballot initiative for the 1998 election must be supported by 54,242 valid signatures, pursuant to Colo. Const. art. V, § 1(2). Montero did not address whether an initiative might reach the ballot without meeting the constitutionally mandated signature requirement. The record before us indicates that in this case the requirement has not been met and Chilcutt’s reasoning would require the Secretary to certify for the ballot a petition that, on the record before us, fails to meet constitutional requirements. Such a result raises more than “technical construction” concerns.
In addition, Fabec and Montero both recognize the need “to fairly guard against ... mistake in the exercise by the people of this constitutional right.” Fabec, 922 P.2d at 341 (quoting Montero, 795 P.2d at 245). Chilcutt views the right of initiative solely from the perspective of an initiative’s proponent who desires to place a measure on the ballot. He ignores the interests of the voters who, through the constitution, have required that an initiative must demonstrate a certain level of support before it may appear on the ballot. To allow certification without a showing that the valid signature requirement has been met would be to require Colorado’s voters to decide on an initiative that has not met a basic constitutional requirement for placement on the ballot. We conclude that this result would fail to protect the integrity of the right of initiative contemplated by our constitution.
C. Statutory Requirements
Turning now to the statute, we apply traditional principles of statutory construction. We initially rely on the language of the statute, giving words and phrases their plain and ordinary meaning. See Moody v. Corsentino, 843 P.2d 1355, 1370 (Colo.1993); People v. Guenther, 740 P.2d 971, 975 (Colo.1987). If explicit statutory provisions are ambiguous or silent regarding the matter at issue, we interpret the statute to comport with the legislature’s objectives. See Brock v. Nyland, 955 P.2d 1037, 1040 (Colo.1998). “A court’s primary task in statutory construction is to ascertain and give effect to the legislative purpose underlying a statutory enactment.” Colorado Common Cause v. Meyer, 758 P.2d 153, 160 (Colo.1988). To do so, we construe the statute as a whole “to give consistent, harmonious, and sensible effect to all of its parts.” AviComm, Inc. v. Colorado Pub. Utils. Comm’n, 955 P.2d 1023, 1031 (Colo.1998). As part of our inquiry we also will consider the legislative history where instructive, see § 2-4-203(c), 2 C.R.S. (1998); Colorado Ass’n of Pub. Employees v. Department of Highways, 809 P.2d 988, 991-92 (Colo.1991), and the potential consequences of a particular construction. See § 2-4-203(e), 2 C.R.S. (1998); State Eng’r v. Castle Meadows, Inc., 856 P.2d 496 (Colo.1993).
The statutory question raised by this case is what consequence follows when the Secretary issues a finding of insufficiency within the thirty days prescribed by section 1-40-116(2), and the random sample on which the finding is based is proven to be erroneous in part. The statute allows a protest to challenge the random sample but does not ad[118]*118dress this circumstance. See § 1-40-118. Chileutt asserts that, given the lack of explicit authority, the petition must be deemed sufficient as a matter of law just as it would if the'Secretary had failed to act within thirty days. See § 1-40-118(1). Chileutt reasons that the Secretary has no authority to conduct a line-by-line count outside the explicit terms of the initiative and referendum procedures defined in the statute.
The Secretary argues that she is authorized by section 1-40-116(4) to conduct the line-by-line review of the signatures. This section delineates the actions she is required to take depending on the random sample results. The Secretary is correct that subsection 116(4) directs her to conduct a line-by-line review of the signatures if the random sample indicates that more than 90% but less than 110% of the required signatures have been submitted.
According to Chileutt, the opportunity to conduct a line-by-line review exists only within the thirty day period for issuing a statement regarding the petition’s sufficiency. Under Chileutt’s analysis, once the thirty-day period for the Secretary’s petition review has passed, “if she does not complete it — for whatever reason — the petition shall be deemed sufficient,” citing section 1-40-118(1). The protest provision does not, however, support this conclusion. Section 118(1) provides in relevant part:
A protest in writing ... may be filed in the district court for the county in which the petition has been filed by some registered elector, within thirty days after the secretary of state issues a statement as to whether the petition has a sufficient number of valid signatures, which statement shall be issued no later than thirty calendar days after the petition has been filed. If the secretary of state fails to issue a statement within thirty calendar days, the petition shall be deemed sufficient.
This section requires the Secretary to issue a statement regarding sufficiency “no later than thirty calendar days after the petition has been filed.” Here the Secretary issued a statement within thirty calendar days. While the random sample proved to be flawed, the finding that the signatures were constitutionally insufficient did not change. There is no requirement that the Secretary both complete the random sample and conduct a line-by-line review within thirty days. To create such a requirement would place a heavy burden on the Secretary and effectively eliminate the use of random sampling. Given the statutory time frame, any time the Secretary’s random sample was shown to be flawed, the initiative would be certified to the ballot. This would create an absurd result which the legislators could not have intended.
An example illustrates the unintended bad consequences of Chileutt’s approach. If, based on a random sample pursuant to section 116(4), the Secretary issues a statement that the petition is presumed to be sufficient, an initiative opponent would have thirty days to protest, just as the proponents had in this case. See § 1 — 10-118(1). Assume that an opponent exercises that right and proves that the random sample was flawed, such that the corrected sample showed that the signatures were either insufficient or within the range calling for a line-by-line count. If the thirty days for the Secretary’s statement have lapsed, as would in all likelihood be the case, then Chileutt’s analysis would require the Secretary to certify the initiative to the ballot. The operation of Chileutt’s approach would render meaningless an initiative opponent’s statutory right to challenge a statement of sufficiency because an opponent, by definition, seeks to prevent an initiative from appearing on the ballot.
For these reasons we reject Chileutt’s analysis and conclude that the Secretary satisfied the thirty day requirement when, within thirty days, she conducted the random sample and issued a statement determining the petition to be insufficient.
Chileutt also argues that placing the initiative on the ballot in these circumstances is consistent with the “statutory tilt” allowing liberal exercise of the initiative right. Protection of the right of initiative is certainly paramount. Article V of the Colorado Constitution reserves initiative power in the people, making it a fundamental right. See Loonan v. Woodley, 882 P.2d 1380, 1383 (Colo.1994); McKee v. City of Louisville, 200 Colo. 525, 530, 616 P.2d 969, 972 (1980). [119]*119Statutory provisions regarding the initiative process should receive a liberal construction to facilitate and not hamper this right. See Fabec v. Beck, 922 P.2d 330, 341 (Colo.1996); Committee for Better Health Care for All Colo. Citizens v. Meyer, 830 P.2d 884, 893 (Colo.1992). This right does not, however, exist without bounds. The breadth of our statutory construction must be limited when “necessary to fairly guard against fraud and mistake in the exercise by the people of this constitutional right.” Id. (internal quotation omitted) (emphasis added). Furthermore, “legislation enacted to facilitate the carrying out of the provisions of the Constitution ... may not avoid or restrict the minimum requirements set out in the Constitution.” Yenter v. Baker, 126 Colo. 232, 239, 248 P.2d 311, 315 (1952) (emphasis added).
The legislative history of the right of initiative is instructive. Cf. § 2-4-203(c), 2 C.R.S. (1998); Colorado Ass’n of Pub. Employees, 809 P.2d at 991-92 (reviewing for legislative intent the historic development of Colorado’s civil service law which, analogous to ballot initiative law, began as a constitutional provision and was “further developed statutorily”). The evolution of both the constitutional language and the initiative legislation illustrates a developing concern with determining the validity of petition signatures.
As originally enacted, the constitutional right of initiative and referendum required signatures by “qualified electors.” See Initiative and Referendum Act, eh. 3, 1910 Colo. Sess. Laws 11 (extraordinary session). That term referred to persons who were eligible to vote; signatories were not required to be registered voters. In 1913, the newly created statutory initiative process included a provision that allowed a petition to be filed with the secretary of state subject to protest by a qualified elector. See Initiative and Referendum Act, ch. 97, sec. 3, 1913 Colo. Sess. Laws 310, 311. Unless a protest was filed, a petition was deemed sufficient. See id. In 1941, the legislature expanded statutory protest opportunities by allowing amended protests if the Secretary denied an initial protest. See Initiative and Referendum Act, ch. 147, sec. 5,1941 Colo. Sess. Laws 480, 485.
In 1980, voters amended article Y, section 1(2) of the constitution by lowering the requisite number of signatures to certify an initiative to the ballot but requiring that the signatures be those of registered electors. See Senate Concurrent Res. No. 7, sec. 1, art. V, § 1(2), (3), 1979 Colo. Sess. Laws 1672, 1672-73 (effective upon proclamation of the Governor, Dec. 19, 1980). As noted above, prior to this amendment, any qualified elector could sign a petition. See id. The reduction in the number of signatures, combined with the heightened voter registration requirement, permitted the verification of signatures. Whereas it was nearly impossible under the “qualified elector” requirement to determine that a signatory was not eligible to vote at the time of signing the petition, voting lists can determine whether a signatory was. a registered elector when the person signed the petition.
The legislature revised the entire process in 1989 and required the Secretary to “examine each name signature on the petition” for any petition submission. See Initiative and Referendum Act, ch. 42, see. 7, § 1-40-109(b)(1), 1989 Colo. Sess. Laws 319, 325. The additions of the- registered elector and signature examination requirements, making signature review practicable and mandatory, have heightened the legal significance of valid petition signatures.
The last significant revision of the initiative and referendum law occurred in 1993, and the legislative history for this, most recent revision is instructive as to the General Assembly’s intent. See, e.g., City of Aspen v. Meserole, 803 P.2d 950, 953-955 (Colo.1990) (relying on. legislative history to determine General Assembly’s intent regarding ambiguous section of the Colorado Governmental Immunity Act). Committee hearings related to the 1993 revision indicate that legislators sought to accomplish two overarching goals: 1) to address continuing concerns about fraud; and 2) to reduce taxpayer costs by streamlining administrative procedures. See Hearings on SB 93-135 Before the Comm. on Bus., 59th Gen. Assembly, 1st Reg. Sess. (Colo.1993) (statement of Rep. Fleming).
The 59th General Assembly viewed random sampling as an effective way to reduce [120]*120fraud and to control administrative costs. See id. There is no indication in the bill’s history that the General Assembly wished to compromise accuracy. In fact, the legislators indicated that, given the growing number of initiatives, random sampling could serve to increase both efficiency and accuracy. See id.; cf. Fabec, 922 P.2d at 340 (Administrative regulation that “assures compliance with the constitutionally prescribed minimum number of [signatures]” is within Secretary’s rulemaking authority.). This history, particularly the important oversight responsibility that the legislature has vested in the Secretary, while not conclusive, contributes to our understanding that accuracy has been and continues to be a significant legislative concern with respect to signature validity.
Just as we may look back to the legislative history for guidance, so may we look forward to the potential consequences of a particular construction. See § 2-4-203(e), 2 C.R.S. (1998); State Eng’r, 856 P.2d at 504. Chil-cutt suggests that if the Secretary- is required to conduct a line-by-line count in the present circumstances, “[i]t would disrupt and jumble the procedural process for certification.” Furthermore, he warns that we would “open the door to a subsequent series of rolling judicial protests and administrative remands,” undermining the need for certainty and finality. We disagree with his prognosis.
With respect to achieving finality, our ruling does not open the door to any administrative or judicial procedures that do not already exist by explicit legislation. In general, the Secretary’s line-by-line count is final but subject to cure or protest. The General Assembly provided initiative proponents with the right to cure deficient signatures by filing an addendum containing additional signatures, see § 1^40-117(b), and allowed any protestor the right to file a protest and have a hearing. See §§ 1^40-118, -119. The cure option does not apply to Chilcutt for the 1998 election because, as explained supra at 115-116, a cure must be completed at least three months before the election. Based on the timing of the submission of this initiative, Chilcutt could not meet that deadline.
Our holding requires a step, prior to cure or protest, that would have occurred but for human error. There is no dispute that, based on the valid signature numbers presented to us by both the petitioner and the respondents, the statute requires a line-byline verification. See § 1-40-116(4). The Secretary erred in undercounting the random sample, but even the corrected number of valid signatures was not sufficient to avoid a line-by-line assessment of each signature. We now require the Secretary to follow the steps she would have taken if she had not made the initial error. This result seems to us to offer certainty that the carefully constructed constitutional and statutory verification requirements, rather than human fallibility, will govern the ballot initiative process. Cf. Fabec, 922 P.2d at 339 (requiring line-byline count after valid signatures on petitioner’s addendum combined with initial random sample to indicate petition contained more than 90% but less than 110% of required total).
For the reasons discussed above, we reverse the district court’s decision ordering the Secretary to certify the Medical Use of Marijuana initiative to the ballot in the 1998 general election. Furthermore, we disapprove of the Secretary’s proposal to certify this initiative to the ballot while she conducted a line-by-line signature review. The proper course would have been to conduct the line-by-line review and, if the signatures had been found sufficient, certify the initiative to the ballot for the year 2000 election. This approach is consistent with the statute’s explicit “Signatures Requirement”:
No petition for any initiated law or amendment to the state constitution shall be of any force or effect, nor shall the proposed law or amendment to the state constitution be submitted to the people of the state of Colorado for adoption or rejection at the polls ... unless the petition ... is signed by the number of electors required by the state constitution.
[121]*121§ 1-40-109(1), 1 C.R.S. (1998) (emphasis added).2
We also find that to order the Secretary to certify the initiative to the ballot without conducting a line-by-line count would create unfortunate incentives within the initiative process. It is quite clear from this ease and other recent cases that initiative petitions generally contain some invalid signatures. Initiative proponents who may not have a sufficient margin for the total signatures to meet the constitutional requirement would be encouraged to file at the last possible date and then challenge a finding of insufficiency on the last possible day. Their hope would be that an error by the Secretary would compensate for their having failed to meet the constitutionally mandated signature requirement.3 The result would be that whenever the Secretary made a good faith error in finding a petition to contain 90% or fewer valid signatures based on the random sample, the proponents effectively could reduce their valid signature requirement by 10%. That is, an initiative could reach the ballot by attaining as little as 90% of the constitutionally required number. This result would conflict with the basic constitutional signature requirement, and we conclude that the Secretary’s good faith error cannot stand as the basis to place on the ballot a matter that did not achieve the constitutionally requisite number of petition signatures.4
Finally, Chilcutt argues that to require a line-by-line count in this situation would create “an exception not made by the Legislature.” Martin v. Montezuma-Cortez Sch. Dist. RE-1, 841 P.2d 237, 246 (Colo.1992)
(quoting Karoly v. Industrial Comm’n, 65 Colo. 239, 245, 176 P. 284, 286 (1918)). We find the opposite to be true. On the record before us, even taken in a light most favorable to the respondents, the medical use of marijuana petition has 52,312 valid signatures or 96.4% of the constitutionally mandated 54,242 signatures. The respondents have' not met the constitutional minimum. To affirm the district court’s order, as Chil-cutt argues we must do, would create an exception to a statutorily and constitutionally mandated signature requirement that the Legislature neither made nor intended.
Consequently, we hold that if, based on a random sample, the Secretary issues a good faith determination of insufficiency in compliance with section 1-40-117, and a timely protest pursuant to section 1-40-118 establishes that the petition contains more than 90% but less than 110% of the required number of valid signatures, then the Secretary is required to conduct a line-by-line examination of each signature. The results of this count are subject to the protest and appeal process ordinarily available under the statute after a line-by-line count.
III. Conclusion
For the reasons discussed above, we find that the Secretary was required to conduct a line-by-line count of the submitted signatures. As we ordered on October 5, 1998, the case has been remanded to the district court for further proceedings consistent with this opinion.
[122]*122• Justice SCOTT concurs in part and concurs in the judgment.
Justice MARTINEZ dissents, and Justice KOURLIS joins in the dissent.