¶1 — Bertha Bashaw was convicted of three counts of delivery of a controlled substance. Because the jury determined that each offense occurred within 1,000 feet of a school bus route stop, her maximum sentence was doubled by statute. Bashaw argues that distance measurements of a mechanical device were improperly admitted because the State failed to demonstrate that the device functioned reliably. Bashaw further contends that the jury instructions incorrectly required unanimity for a finding that her actions did not take place within 1,000 feet of the school bus route stop. We agree with both of Bashaw’s arguments, though we find the improper admission of the results harmless with respect to two of the sentence enhancements. Because the instructional error was not harmless, however, we reverse all three sentence enhancements and remand the case to the trial court for further proceedings consistent with this opinion.
FACTS
¶2 In July 2007, the State charged Bashaw with three counts of delivery of a controlled substance based on three separate sales to a police informant. The State also sought a sentence enhancement, pursuant to RCW 69.50.435(l)(c), based on its allegation that each sale took place within I, 000 feet of a school bus route stop. That enhancement allows for imprisonment of up to twice the period otherwise authorized. Count I was alleged to have occurred on May II, 2006, some distance south of the former Vaagen Mill’s parking lot. Counts II and III were alleged to have occurred on May 23 and May 31, 2006, respectively, in the parking lot of the former mill.
¶3 At trial, witness testimony established the locations of the school bus route stops and the drug transactions. Dan Chaplik, superintendent of the Republic School District, testified to the locations of two school bus route stops in the area. One was in the “main driveway” of the old mill site, which was located in the parking lot, while the other was *138across the street and slightly to the south of the parking lot. 1 Transcript of Trial (TR) at 53, 56. Detective Jan Lewis testified that he returned to the locations of the transactions and measured the distance from each location to the nearest school bus route stop. All three drug transactions took place in the vicinity of the same two school bus route stops. To measure the distances, Detective Lewis used what he described as “[o]ne of those rolling wheel measurers you can zero out and roll along ahead of you and it counts out feet.” 2 TR at 176. Detective Lewis further testified that he borrowed the particular device from the Republic Police Department and that he had not used it before, though he had used similar devices. Such devices, according to Detective Lewis, are commonly used by law enforcement. After Detective Lewis pressed a button to zero out the numbers, he measured the distance from each transaction location to the school bus route stop.1 At trial, Bashaw objected to the admission of the results of the measuring device based on a lack of foundation. The trial judge overruled the objection, and Detective Lewis testified that, based on his measurements, the distance from the location of the first sale to the school bus route stop was 924 feet and the distances from the locations of the second and third transactions to the school bus route stop were each 100 to 150 feet. This was the only testimony directly addressing the distance between the transactions and the school bus route stop.
¶4 The transcript also reveals additional testimony about the relevant locations from which distance might be inferred. Detective Lewis estimated that the distance from the parking lot entrance to the end of the former mill’s parking lot was no more than 150 feet. Additionally, four other witnesses, including the confidential informant and three other law enforcement personnel, testified to the *139locations of the drug transactions and the distance from the parking lot to the location of the transaction alleged in count I. As to this distance, Detective Donald Redfield estimated the distance to be one-tenth of a mile (528 feet) while Detectives Steve Brown and Armondo Moralez, as well as the confidential informant, estimated the distance to be one-quarter of a mile (1,320 feet).
¶5 Because the State sought a sentence enhancement, the jury was given a special verdict form for each charge. The form asked the jury to make a special finding of whether each charged delivery took place within 1,000 feet of a school bus route stop. In the jury instruction explaining the special verdict forms, jurors were instructed, “Since this is a criminal case, all twelve of you must agree on the answer to the special verdict.” Clerk’s Papers at 95. On appeal, Bashaw challenges this instruction as contrary to precedent from this court.
¶6 The jury found Bashaw guilty of all three counts of delivery of a controlled substance and found that each had taken place within 1,000 feet of a school bus route stop. The latter finding increased Bashaw’s maximum sentence from 24 months to 48 months. The trial judge sentenced Bashaw to 36 months’ imprisonment. Bashaw appealed the sentence but not the underlying conviction. The Court of Appeals affirmed the sentence, and Bashaw filed a petition for review with this court, which we granted. State v. Bashaw, 144 Wn. App. 196, 182 P.3d 451, review granted, 165 Wn.2d 1002, 198 P.3d 512 (2008).
ISSUES
¶7 1. Did the trial court abuse its discretion by admitting testimony about the results of a measuring device without any showing of reliability?
¶8 2. Did the trial court correctly instruct the jury that its special finding had to be unanimous?
*140STANDARD OF REVIEW
¶9 This court reviews challenged jury instructions de novo. State v. Bennett, 161 Wn.2d 303, 307, 165 P.3d 1241 (2007). We review the admission of evidence for abuse of discretion. City of Auburn v. Hedlund, 165 Wn.2d 645, 654, 201 P.3d 315 (2009). “Abuse of discretion exists ‘[w]hen a trial court’s exercise of its discretion is manifestly unreasonable or based upon untenable grounds or reasons.’ ” State v. Magers, 164 Wn.2d 174, 181, 189 P.3d 126 (2008) (alteration in original) (quoting State v. Powell, 126 Wn.2d 244, 258, 893 P.2d 615 (1995)).
ANALYSIS
I. The Trial Court Abused Its Discretion by Admitting Testimony about the Results of a Measuring Device without Any Showing of Reliability
¶10 The first issue in this case concerns the showing of reliability necessary for a trial court to admit testimony about the results of a measuring device. In accordance with analogous precedent, we hold that admission of results from a distance measuring device requires a showing that the particular device was functioning properly and produced accurate results. Because the State produced no evidence that the distance measuring device here produced accurate results, its admission was error and an abuse of discretion. That error, however, was harmless as to counts II and III but not as to count I. Accordingly, we vacate the sentence enhancement with respect to count I on this basis.
A. Evidence Must Be Authenticated Prior to Admission
¶11 It is fundamental that evidence must be authenticated before it is admitted. See ER 901(a). Authentication requires that the proponent produce proof “sufficient to support a finding that the matter in question is what its proponent claims.” Id. The party offering the evidence must *141make a prima facie showing consisting of proof that is sufficient “to permit a reasonable juror to find in favor of authenticity or identification.” State v. Payne, 117 Wn. App. 99, 106, 69 P.3d 889 (2003); see also Judicial Council cmt. 901, cited in 5C Karl B. Tegland, Washington Practice: Evidence Law and Practice § 901.1, at 283 n.3 (5th ed. 2007).
¶12 Conceptually, authentication is a process of establishing conditional relevance. See Judicial Council cmt. 901, cited in 5C Tegland, supra, § 901.1, at 283 n.3; see also Robert H. Aronson, The Law of Evidence in Washington § 901.05(1), at 901-12 (4th ed. 2008) (“Unless evidence is in fact what it purports to be, it is not relevant.”). As observed in Washington Practice, “a photograph might be relevant, but only if it accurately depicts the subject”; “[an audio] recording might be relevant, but only if the sounds were recorded faithfully and the voices are accurately identified.” 5C Tegland, supra, § 901.1, at 283. Likewise, a distance measurement may be relevant, but only if it is accurately measured.
¶13 In a line of cases analogous to the present one, the courts of this state have held that, under ER 901, speed measuring devices, such as radar devices, must be authenticated in order for their results to be admissible. See City of Bellevue v. Mociulski, 51 Wn. App. 855, 859-60, 756 P.2d 1320 (1988); see also City of Bellevue v. Hellenthal, 144 Wn.2d 425, 431-32, 28 P.3d 744 (2001) (citing Mociulski, 51 Wn. App. at 860-61, with approval); City of Bellevue v. Lightfoot, 75 Wn. App. 214, 221, 877 P.2d 247 (1994) (“police traffic radar results are not admissible unless the particular radar device used is shown to be reliable”); City of Seattle v. Peterson, 39 Wn. App. 524, 527, 693 P.2d 757 (1985) (holding that evidence of a machine’s reliability is a prerequisite to admission of the machine’s results). Authentication of such devices requires a showing that the particular unit “was functioning properly and produced accurate *142results” at the time it was employed. Lightfoot, 75 Wn. App. at 221.2
¶14 We agree with the formulation of the Court of Appeals, as expressed in the speed measuring device line of cases, regarding the authentication required prior to admission of measurements made by mechanical devices.3 The rules of evidence, analogous case law, and common sense all dictate that before the State introduces evidence that will result in a mandatory penalty enhancement, the State must show that the evidence it relies upon is accurate. Simply put, results of a mechanical device are not relevant, and therefore are inadmissible, until the party offering the results makes a prima facie showing that the device was functioning properly and produced accurate results. This is consistent with the rationale underlying the requirement of authentication. See 5C Tegland, supra, § 901.1, at 283. As such, we hold that the principle articulated in the context of speed measuring devices also applies to distance measuring devices: a showing that the device is functioning properly and producing accurate results is, under ER 901(a), a prerequisite to admission of the results.
¶15 It is true, of course, that electronic instruments differ from standard rolling wheel measuring devices in complexity. That difference, however, is properly addressed through what prima facie showing is required rather than whether a prima facie showing is required.4 In the present case, the State failed to make a prima facie showing that the rolling wheel measuring device produced accurate re-*143suits. Though we know that the device displayed numbers and that it “click [ed] off feet and inches” while Detective Lewis pushed it, no testimony or evidence even suggested that those numbers were accurate. 2 TR at 181. No comparison of results generated by the device to a known distance was made nor was there any evidence that it had ever been inspected or calibrated. The trial court abused its discretion by admitting the results of the rolling wheel measuring device with no showing whatsoever that those results were accurate.
B. Improper Admission of Evidence Was Harmless as to Counts II and III but Not as to Count I
¶16 The improper admission of evidence to support a criminal conviction may be harmless error. State v. Flores, 164 Wn.2d 1, 18, 186 P.3d 1038 (2008). An evidentiary error is not harmless “if, ‘within reasonable probabilities, had the error not occurred, the outcome of the trial would have been materially affected.’ ” State v. Neal, 144 Wn.2d 600, 611, 30 P.3d 1255 (2001) (quoting State v. Smith, 106 Wn.2d 772, 780, 725 P.2d 951 (1986)). The record allows us to conclude that the improper admission of the results of the rolling wheel measuring device would not have materially affected the jury’s special verdicts with respect to counts II and III. The outcome of the special verdict with respect to count I, however, might well have been different had the court excluded the results.
¶17 Apart from the measurements made by the rolling wheel measuring device, testimony at trial also addressed the relevant locations and distances between the drug transactions and a school bus route stop. Superintendent Chaplik testified, and, based on the jury’s verdicts, it must have believed, that the school bus stopped in “the driveway area of the Vaagen’s Mill site,” also referred to as the parking lot. 1 TR at 53. Numerous witnesses, including law enforcement officers and the confidential informant, testified that the transactions that were the basis for counts II and III took place in the parking lot. Id. at 110,136; 3 TR at *144275-77, 284-86, 322, 344-45, 358-60; 4 TR at 400, 405. Again, to reach its conclusions, the jury must necessarily have found the testimony about these locations credible. Detective Lewis estimated that the parking lot, in which the bus stopped and the transactions occurred, is no more than 150 feet long. 2 TR at 183. In the presence of such extensive testimony, much of which this jury would necessarily have found credible to reach its verdict, we conclude that there is no reasonable probability that this jury, as instructed, would have concluded that the special verdicts relating to counts II and III did not take place within 1,000 feet of a school bus route stop if the results of the rolling wheel measuring device had been excluded.
¶18 As to count I, there is a reasonable possibility that the jury would have reached a different conclusion on the special verdict if the improperly admitted results had been excluded. After meeting at the parking lot of the mill, Bashaw and the confidential informant traveled some distance south to conduct the transaction. One detective estimated the distance was one-tenth of a mile, or 528 feet, from the parking lot entrance, 3 TR at 275, while two other detectives and the confidential informant estimated the distance was one-quarter of a mile, or 1,320 feet, id. at 314, 339; 4 TR at 391. Thus, the testimony was conflicting but weighed in favor of finding that the distance was over 1,000 feet from the parking lot. Though an aerial photograph of the area was entered into evidence, it contained no scale or other method of accurately determining distance. The photograph does not establish that the transaction occurred within 1,000 feet of a school bus route stop. As to the special verdict on count I, then, there is at least a reasonable probability that excluding the results of the rolling wheel measuring device would have materially affected the outcome. As such, the improper admission of the results of the rolling wheel measuring device was not harmless and the special verdict with respect to count I must be vacated.
*145II. The Trial Court Incorrectly Instructed the Jury That Its Special Finding Had To Be Unanimous
¶19 The jury instruction issue in this case is a narrow one: when a jury has unanimously found a defendant guilty of a substantive crime and proceeds to make an additional finding that would increase the defendant’s sentence beyond the maximum penalty allowed by the guidelines, must the jury’s answer be unanimous in order to be final? We answered this question in State v. Goldberg, 149 Wn.2d 888, 72 P.3d 1083 (2003), and the answer is no. Anonunanimous jury decision on such a special finding is a final determination that the State has not proved that finding beyond a reasonable doubt.5
¶20 In Goldberg, the defendant was charged with first degree murder, pursuant to RCW 9A.32.030, with an aggravating circumstance enumerated in RCW 10.95.020.149 Wn.2d at 893. The finding of an aggravating circumstance would have increased the maximum penalty to “life imprisonment without possibility of release or parole.” RCW 10.95.030(1). The jury in Goldberg initially returned a verdict finding the defendant guilty of first degree murder but answered “no” on the special verdict form asking whether the aggravating circumstance was present. 149 Wn.2d at 891. The judge polled the jury and found that one juror had voted “no” on the aggravating factor.6 Id. The presiding juror informed the judge that there was no reasonable probability of the jury reaching a unanimous agreement within a reasonable time. Id. Despite that, the judge ordered the jury to continue deliberations the next day and the jury subsequently returned a unanimous finding that the State had proved the aggravating factor. Id. at 891-92.
*146¶21 In resolving the appeal in Goldberg, we rejected the parties’ framing of the issue as one of jury coercion. Id. at 893. Instead, the issue we addressed was “whether . . . unanimity is required” for a special finding increasing the maximum penalty and we held that “it is not.” Id. We went on to hold that the “jury’s [nonunanimous] judgment should have been accepted” and that it was error to order continued deliberations. Id. at 894. We concluded by stating, “In sum, special verdicts do not need to be unanimous in order to be final.” Id. at 895. The rule from Goldberg, then, is that a unanimous jury decision is not required to find that the State has failed to prove the presence of a special finding increasing the defendant’s maximum allowable sentence. A nonunanimous jury decision is a final determination that the State has not proved the special finding beyond a reasonable doubt.7
¶22 The rule we adopted in Goldberg and reaffirm today serves several important policies. First, we have previously noted that “[a] second trial exacts a heavy toll on both society and defendants by helping to drain state treasuries, crowding court dockets, and delaying other cases while also jeopardizing the interests of defendants due to the emotional and financial strain of successive defenses.” State v. Labanowski, 117 Wn.2d 405, 420, 816 P.2d 26 (1991). The costs and burdens of a new trial, even if limited to the determination of a special finding, are substantial. We have also recognized a defendant’s “ Valued right’ to have the charges resolved by a particular tribunal.” State v. Wright, 165 Wn.2d 783, 792-93, 203 P.3d 1027 (2009) (internal quotation marks omitted) (quoting Arizona v. Washington, 434 U.S. 497, 503, 98 S. Ct. 824, 54 L. Ed. 2d 717 (1978)). Retrial of a defendant implicates core concerns of judicial economy and finality. Where, as here, a defendant is already subject to a penalty for the underlying substantive *147offense, the prospect of an additional penalty is strongly outweighed by the countervailing policies of judicial economy and finality.
¶23 Applying the Goldberg rule to the present case, the jury instruction stating that all 12 jurors must agree on an answer to the special verdict was an incorrect statement of the law. Though unanimity is required to find the presence of a special finding increasing the maximum penalty, see Goldberg, 149 Wn.2d at 893, it is not required to find the absence of such a special finding. The jury instruction here stated that unanimity was required for either determination. That was error.
¶24 In order to hold that a jury instruction error was harmless, “we must ‘conclude beyond a reasonable doubt that the jury verdict would have been the same absent the error.’ ” State v. Brown, 147 Wn.2d 330, 341, 58 P.3d 889 (2002) (quoting Neder v. United States, 527 U.S. 1, 19, 119 S. Ct. 1827, 144 L. Ed. 2d 35 (1999)). The State argues, and the Court of Appeals agreed, that any error in the instruction was harmless because the trial court polled the jury and the jurors affirmed the verdict, demonstrating it was unanimous. This argument misses the point. The error here was the procedure by which unanimity would be inappropriately achieved. In Goldberg, the error reversed by this court was the trial court’s instruction to a non-unanimous jury to reach unanimity. 149 Wn.2d at 893. The error here is identical except for the fact that that direction to reach unanimity was given preemptively.
¶25 The result of the flawed deliberative process tells us little about what result the jury would have reached had it been given a correct instruction. Goldberg is illustrative. There, the jury initially answered “no” to the special verdict, based on a lack of unanimity, until told it must reach a unanimous verdict, at which point it answered “yes.” Id. at 891-93. Given different instructions, the jury returned different verdicts. We can only speculate as to why this might be so. For instance, when unanimity is required, jurors with reservations might not hold to their positions or *148may not raise additional questions that would lead to a different result. We cannot say with any confidence what might have occurred had the jury been properly instructed. We therefore cannot conclude beyond a reasonable doubt that the jury instruction error was harmless. As such, we vacate the remaining sentence enhancements and remand for further proceedings consistent with this opinion.
CONCLUSION
¶26 We reach two conclusions in this case. First, testimony about the results of a mechanical device is admissible only if there is some showing that the particular measuring device was functioning properly and producing accurate results. Second, a nonunanimous special finding by a jury is a final decision by the jury that the State has not proved its case beyond a reasonable doubt. Because there was no showing that the distance measuring device employed here produced accurate results and that error was not harmless as to the special verdict on count I, and because the trial court erred in instructing the jury on the unanimity requirements for special findings, we reverse all three sentence enhancements and remand the case to the trial court for further proceedings consistent with this opinion.
C. Johnson, Sanders, Chambers, Fairhurst, and Stephens, JJ., concur.The State has conceded on appeal that all three distances were the result of measurement with the same device. Wash. Supreme Court oral argument, State v. Bashaw, No. 81633-6 (Sept. 17,2009), at 21 min., 18 sec., video recording by TVW, Washington State’s Public Affairs Network, available at http://www.tvw.org; Resp’t’s Br. at 3 (adopting Bashaw’s statement of the case, including her assertion that all three distances were measured, Appellant’s Br. at 5).
For speed measuring devices, this showing is now governed by CrRLJ 6.13 and IRLJ 6.6. No comparable rule exists for distance measuring devices.
For devices relying on complex scientific principles, authentication is actually a compound determination, involving first the qualifications of a witness under ER 702 and then whether the mechanical device operated reliably. Hellenthal, 144 Wn.2d at 432. No ER 702 question is before the court in this case.
Some devices operate in a maimer such that any failure by the device to produce accurate results would be immediately obvious to the user (e.g., measuring tapes, yard sticks, or rulers). In such cases, it may be inferred from testimony by the user about measurements with the device that the results are accurate. 'Ibis contrasts with rolling wheel measuring devices for which, like speed measuring devices, the internal workings are not observable by the user.
General verdicts in criminal cases, of course, must still be unanimous to convict or acquit. See Wash. Const, art. I, § 21; State v. Stephens, 93 Wn.2d 186, 190, 607 P.2d 304 (1980).
In fact, three jurors had voted "no” but only one juror raised a hand when asked. Goldberg, 149 Wn.2d at 891.
This rule is not compelled by constitutional protections against double jeopardy, cf. State v. Eggleston, 164 Wn.2d 61, 70-71, 187 P.3d 233 (stating that double jeopardy protections do not extend to retrial of noncapital sentencing aggravators), cert. denied, 555 U.S. 1075 (2008), but rather by the common law precedent of this court, as articulated in Goldberg.