Dimensional Review of Scales for Forensic Photography | NIST, Study notes of Photography

Download Dimensional Review of Scales for Forensic Photography | NIST and more Study notes Photography in PDF only on Docsity! Page | 1 Dimensional Review of Scales for Forensic Photography Massimiliano Ferrucci1, Theodore D. Doiron1, Robert M. Thompson2, John P. Jones II2, Susan M. Ballou2, Janice A. Neiman3 1Dimensional Metrology Group, 2Law Enforcement Standards Office National Institute of Standards and Technology (NIST), 3University of Delaware Abstract Scales for forensic photography provide a geometrical reference in the photographic documentation of evidence. A common scale used by investigators is a plastic, L- shaped ruler that allows for a dimensional reference in the photographic documentation of evidence or a crime scene. The ABFO No. 2 Standard Reference Scale has received recognition by the forensic science community as a reliable and accurate reference scale; however, we have surveyed commercially available scales and found a lack of consistency in manufacturing processes and, consequently a lack of strict adherence to the standard. This study seeks to evaluate the quality of commercially available photo scales, document manufacturing processes, and suggest pathways for establishing standards for forensic photo scales that will serve as a means for ensuring accuracy and user confidence. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 2 Table of Contents I. Introduction a. ISO 17025 accreditation of forensic laboratories b. The ABFO No. 2 Standard Reference Scale c. Concerns over accuracy II. Methods III. Results a. Initial testing b. Four-year revisit and measurement c. Manufacturer visits d. Surveillance testing IV. Conclusions a. Current industry standards b. Suggestions for a forensic photo scale standard V. Notes This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 5 Hyzer and Krauss defined a set of specifications for the design of the ABFO No. 2.4 The specifications dictate the shape, size, position and chromatic attributes of printed features. For this study, we are investigating those specifications of the ABFO No.2 that affect the quantitative (metric) analysis of photographs.5 These are: 1. “The overall accuracy of scale ABFO No. 2 is ± 0.1 mm or ± 1% for the major centimetre graduations.” 2. “The internal and external diameters of the three circles are 19.75 and 23.00 mm, respectively.” 6 3. “The error in placement of the three circles is within 0.25% of the nominal 80- mm separation between their centers.” 4. “The legs are mutually perpendicular to ± 2 min of arc.” 7 Specifications 2, 3, and 4 are depicted in Figure 2 below. Figure 2: ABFO No. 2 specifications dictate the accuracy of the length scale, the perpendicularity of the rulers, the distance between the centers of the circles, and the internal and external diameters of the circles. 4 Hyzer, W. G. and Krauss, T. C., “The Bite Mark Standard Reference Scale – ABFO No. 2,” Journal of Forensic Sciences, JFSCA, Vol. 33, No. 2, March 1988, pp. 498-506. 5 See Discussion section Note 1 on other features not observed in this study. 6 See Discussion section Note 2 on conflicting specifications regarding diameters. 7 Min of arc is also known as arcminute. 1 arcminute = 1/60 degrees. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 6 The ABFO No. 2 was originally developed for documentation of patterned injuries thought to be bite marks; however, its design and subsequent products have been adopted for laboratory and crime scene photography in general. c) Concerns over accuracy The task of producing the ABFO No. 2 was initially granted to one manufacturer. However, a review of commercially available photo scales showed a proliferation of rulers similar in design to the original ABFO No. 2 (Figure 3). In some cases, rulers were advertised as ‘ABFO No. 2 Photomacrographic Scale,’ without the printed ‘ABFO No. 2’ label on the rulers themselves. Figure 3: A review of commercially available photo scales showed a proliferation of rulers similar in design to the ABFO No. 2. The Dimensional Metrology Group (DMG) sought to evaluate the conformance of commercially-available photo scales to the ABFO No. 2 specifications. While the photo scales in the study were chosen for their similarity to the ABFO No. 2 design, it should be noted that not all were advertised as ‘ABFO No. 2.’ This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 7 II. Methods The DMG measured dimensional features from a sample of commercially available photo scales, using as criteria the ABFO No. 2 specifications. Five rulers were purchased from each of four vendors. The following features were inspected on each ruler: (a) the accuracy of the horizontal and vertical length scale graduations, (b) the perpendicularity of the legs, (c) the inner and outer diameters of the circles, and (d) the center-to-center distance between circles. Each feature was measured five times and the average value is reported. Measurements were made on the rulers over several years and on two vision measurement machines. The rulers were purchased in 2007 and were subsequently measured under a Mitutoyo UMAP vision measurement system. The UMAP is equipped with a vision probe capable of edge detection and focusing for coordinate mapping with accuracies of a few micrometers (Figure 4). The rulers were held flat with putty on the extremities (Figure 5) to avoid errors as a result of bending. The DMG used the UMAP to measure features (a) – (c). Figure 4: The UMAP vision measurement system is equipped with a vision probe capable of edge detection and focusing for coordinate mapping with accuracies of a few micrometers. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 10 Equation 2: Vendor’s stated uncertainty for measurements in X and Y plane. (VMR) It should be noted that the measurement uncertainty is a function of the size of the measurand. In this case, the smallest displacement measured was 10 mm, while the largest displacement didn’t exceed 85 mm. Thus, the uncertainty range contributed by each measurement system is given below (Equation 3). Equation 3: Measurement uncertainty for measurement range covered in this study. The error in the measurement systems is almost two orders of magnitude smaller than the tolerance values given in the ABFO No. 2 specifications. The largest source of error would actually be the print quality of features on rulers. Figure 8 gives an example of the variation in graduation print quality between vendors.8 We can, therefore, consider any difference in the performance of the systems to be negligible. Figure 8: The print quality of features, such as the length graduations seen above, contributes the largest error in the measurements. To understand the behavior of photo scales over time, we measured the same rulers in 2011, four years after purchase. All features were re-measured on the Nikon 8 Please see Discussion section, Note 3 for more examples of print quality among rulers. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 11 vision system using the same measurement scheme. The new results are presented and compared to previous data. The DMG investigated the primary sources of forensic photo scales in the United States and found that the production is limited to only a few manufacturers. Site visits were made to these manufacturers in order to learn how they produce forensic photo scales and how they control the quality of their production. The DMG’s findings are discussed. Once the production and distribution of forensic photo scales was better understood, an extensive surveillance of a new, larger sample of rulers was performed. The Nikon VMR-6555 vision measurement system was used for this round of testing. The testing scheme for each feature is consistent with previous tests. However, this time the rulers were not held flat so as to allow us to quantify the effect of bending on the quality of the length scale. III. Results a) Initial testing Length Scales The accuracy of the scale is determined by the error in position of the length graduations with respect to ‘zero.’ The measured distance between the zero graduation and each subsequent graduation is compared to the nominal distance, and the error is extracted. For example, if the position of the 2 cm graduation is a distance of 2.05 cm from the 0 (zero) cm graduation, then its error is +0.05 cm. We define the position of a graduation as the intersection of its midline and the inner edge of the ruler (Figure 9). The specification states that the accuracy of the scale is “± 0.1 mm or ± 1% for the major centimetre graduations.” We understand that to mean that the allowed error between adjacent millimeter graduations is ± 0.1 mm, while the cumulative error at each This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 12 centimeter graduation shall not exceed ± 1% from its nominal value. By investigating the error in position of centimeter graduations should give a reasonable indication of the relative behavior between millimeter graduations. Thus, we measure the error in position for the first five centimeter graduations. Both the vertical and the horizontal scales were tested on each ruler (Figure 10). Figure 9: The position of a graduation is defined as the intersection of the graduation’s midline and the inner edge of the ruler. Note the concentric crosshairs. The intersection of the crosshairs denotes the extracted position. Figure 10: The test measured the error in position of the first five centimeter graduations. Both the vertical and horizontal scales were tested. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 15 External Diameter Vendor Average External Diameter (mm) Standard Deviation (mm) Variability (%) A 20.124 0.007 0.037 B 20.396 0.030 0.145 C 20.379 0.029 0.141 D 19.176 0.015 0.076 Table 2: Measurements on external diameter were made on each of three circles for every ruler. The average external diameter is reported for each vendor, along with the standard deviation and variability as a percentage value from average. The specification for external diameter is given to be 23.00 mm.9 The average values for external diameter indicate that none of the vendors were close to this specification. However, it must be noted that the variability is relatively low within each vendor, thus demonstrating a relatively consistent manufacturing process. Vendor A, again, exhibits the lowest variability among its rulers. Center-to-center distance between circles The measurements of center-to-center distance between circles (and all subsequent measurements in this study) were performed on the Nikon VMR-6555. The first step is to define the position of a circle’s center. The center of a circle can be defined as either (1) the intersecting point of the two concentric crosshairs, or (2) by fitting a circle to the internal or external edges and extracting the calculated center (Figure 12). In the first method, we first find the midline of each crosshair. Then, we take the point of intersection of the two midlines as the center of the circle. Whereas, in the second method, the circle-fitting function provides the user with the point coordinates for the circle’s center. Once we established point coordinates for each of the three circles, we can calculate the distances by using the Pythagorean Theorem (Equation 4). We will perform measurements using both methods and compare the results. 9 See Discussion section on conflicting specifications This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 16 Equation 4: When the point coordinates are known, the distance can be calculated using the Pythagorean Theorem. Δx and Δy are the displacement in x and y coordinates between circle centers. Figure 12: The centers are measured by two methods. The first method assumes the center of the circle is given by the intersection of the concentric crosshairs. The second method uses a circle-fitting function on the circumference of the circle to extract its center. Each ruler has two center-to-center segments (Figure 13). Since the specification is the same for both the horizontal (d1-2) and the vertical (d1-3) segments, we don’t distinguish the two in our results. For each vendor, the average of all distance measurements is presented. The standard deviation and its percentage value with respect to average (variability) are also presented (Table 3). Figure 13: There are two center-to-center segments on each ruler, d1-2 and d1-3. Both are measured and results are not distinguished. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 17 Vendor Crosshairs Average (mm) Standard Deviation (mm) Variability (%) Circle- fitting Average (mm) Standard Deviation (mm) Variability (%) A 79.749 0.024 0.031 79.745 0.022 0.028 B 80.024 0.024 0.030 80.027 0.025 0.031 C 85.099 0.081 0.095 N/A N/A N/A D 77.763 0.077 0.099 77.767 0.062 0.079 Table 3: Measurements of center-to-center distance were made using two methods. Each ruler had two center-to-center segments. The average over five rulers for each vendor is reported, along with the standard deviation and variability as a percentage value from average. We recall that the ABFO No. 2 specification for distance between the centers of the circles is 80 mm ± 0.25% (or ± 0.2 mm). Therefore, any values below 79.8 mm and above 80.2 mm are considered not in conformance. The average values from the ‘crosshair’ and ‘circle-fitting’ method differ by no more than 0.004 mm (4 µm). This is negligible considering that it is in the same order of magnitude as the repeatability of the measurement system. We could not extract circle-fitting data from vendor C since the circumference of the circles had printed obstructions (Figure 14). Hence, we will use the results from the ‘crosshairs’ method. Figure 14: Circle-fitting was not performed on vendor C rulers since part of the circumference was obstructed by other printed features. The results indicate that only vendor B satisfies the ABFO No. 2 specification. However, it should be noted that, despite the discrepancy in center-to-center distance between vendors, the variability within each vendor is extremely low (less than 0.1%). As previously mentioned, this indicates a consistent manufacturing process. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 20 The method of using the center-to-center length segments as the legs and the cosine law had its shortcomings. Small errors in the length measurement of the segments would yield large errors in the estimation of angle. We look at the propagation of error from length measurement to estimation of angle. The law of cosines is rewritten: We can then use the formula for error propagation to calculate the error in due to errors in , , and : We set the length segments, , , and , to a default 80 mm, 80 mm, 113.14 mm, respectively. At the same time, we set their errors, , , and , to 0.01 mm. This is a reasonable error in length measurement given the print quality of the rulers (Figure 18). Figure 18: The error in print quality of the rulers, as shown in this image, produce errors in the determination of the crosshair intersection. This error will then propagate to the calculation of angle when testing a ruler’s perpendicularity. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 21 We find that, given these values, the error in angle calculation comes to Considering that the specified angle tolerance in the ABFO No. 2 technical document is ± arcmins, the error associated with our method of calculating the angle is substantial. However, the error in the alternate method of using the internal edges of the ruler would have been significantly higher, given the lack of straightness and imperfections in the edges. b) Four-year revisit and measurement After several years, we noticed the rulers had developed a grayish yellow stain throughout their surfaces. This is not unexpected, as plastic is not considered a stable material. However, it led us to wonder if the rulers undergo dimensional changes over time. To better understand the behavior of the rulers over a long period of time, we retested the same rulers four years after purchase. The rulers were measured using the Nikon VMR-6555 vision measurement machine. All features except for circle diameters were studied. Length Scale The new data is plotted separately for each vendor (Figure 19). The change is also plotted. Negative change indicates shrinking of the scale, while positive change indicates a growing scale. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 22 This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 25 c) Manufacturer visits Site visits of the manufacturers revealed manufacturing processes and methods of quality control. Depending on the material of the ruler, there are various methods of impressing features. Ink-based digital printing is suitable for thinner, usually paper- based, rulers. Silk screen printing is used for hard plastic rulers, such as the ABFO No. 2. Etching, on the other hand, is often used for steel rulers. A common technique for quality control is to check the printed scales with a ‘certified’ steel ruler. The steel ruler is placed flush against the printed ruler’s edge so that the positions of the graduations can be compared. The units being compared should be consistent (i.e. English with English or metric with metric). The ‘zero’ graduation of the photo scale is lined up with the zero graduation of the steel ruler. In cases that the ‘zero’ of the steel ruler is the edge of the ruler itself (Figure 21), it is suggested to line up the zero graduation of the photo scale with the ‘1’ graduation of the steel ruler. Figure 21: A common method for checking the accuracy of a plastic ruler’s length scale is to compare it to the scale of a steel ruler. Sometimes, the ‘zero’ of the steel ruler’s scale is defined to be the ruler’s edge. In this case, it is suggested that the next graduation (of equal unit) be used as the ‘zero.’ An eyepiece can help in the comparison by providing magnification (Figure 22). Care should be taken to reduce parallax caused by the viewing angle, with or without an eyepiece. It should be noted that manufacturers did not mention checking any other feature other than from the length scale. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 26 Figure 22: An eyepiece can help in comparing two length scales placed flush against each other. Care has to be taken to avoid parallax error caused by oblique viewing angles. When discussing quality control, manufacturers mentioned the use of ‘certified’ steel rulers. Often, a ‘certified’ ruler is understood to mean a ruler calibrated by a laboratory whose length measurements are traceable to a National Metrology Institute (NIST in the United States). At the moment, there are no standard procedures for the testing of forensic photo scales subsequent to manufacture and sale. While a calibrated ruler establishes a certain level of measurement confidence, it is not necessary for checking plastic rulers. Coincidentally, our group previously performed a study of the accuracy of three differently-priced steel rulers using the Nikon machine to make measurements. The results show that, while the more expensive steel rulers had minimal errors, the least expensive steel ruler’s errors were still small when compared to tolerance values for plastic rulers. In this study, the least expensive ruler was off by a little more than 0.1 mm over its entire length of 300 mm, or 0.033% from nominal, which is almost two orders of magnitude smaller than the ABFO No. 2 specification. The data is shown below (Figure 23). This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 27 Figure 23: A study of length scales on steel rulers shows that, while the more expensive rulers had minimal errors, the errors on the least expensive were still much smaller than the errors seen on plastic rulers. Therefore, while the use of certified steel rulers is good practice, it might not be absolutely necessary for the purpose of verifying that photo scales meet the ABFO No. 2 specifications. Manufacturers typically have a documented procedure for deciding whether or not the printed scale is acceptable. If rulers are printed in large quantities, it is typical for the vendor to check a ruler at the beginning of the printing process, one in the middle, and one at the end. When a ruler is checked and found to be outside the manufacturer’s tolerance, the printing is stopped. All rulers printed before the failed check, but after the last successful check, are rejected. Appropriate adjustments are then made to the setup and printing is resumed. When asked about tolerance, manufacturers quoted their own values, without mention of ABFO No. 2 specifications. Also, instead of quoting length- dependent tolerance values, manufacturers quoted tolerances over the entire length of the scale. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 30 Figure 25: The error in position for each centimeter graduation is plotted against its nominal value. The mean error between each centimeter graduation is connected by a solid line. Each vendor is reported separately. The results show that, much like in the previous study, all rulers satisfy the “± 1% for the major centimetre graduations” tolerance set by the ABFO No. 2 technical note. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 31 Center-to-center distance between circles For the center-to-center distance, we used only the intersection of the crosshairs to denote the center of the circle. The data is presented in Table 5 below. Vendor Crosshairs Average (mm) Standard Deviation (mm) Variability (%) A 79.743 0.043 0.053 B 79.641 0.039 0.049 C 77.721 0.046 0.059 D 80.084 0.054 0.068 E 79.738 0.012 0.016 F 84.999 0.165 0.194 G 84.924 0.063 0.075 H 85.179 0.046 0.055 I 84.942 0.233 0.275 J 82.527 0.028 0.034 Table 5: Only the crosshair intersection method was used in the surveillance testing of center-to-center distance. The average over five rulers for each vendor is reported, along with the standard deviation and variability as a percentage value from average. Perpendicularity of legs The law of cosines was used on the three length segments. We note the error in angle calculation due to errors in measurement of distance as previously discussed. Vendors A – J are displayed from top to bottom (Figure 26), with vendor A being at the top (yellow) and vendor J at the bottom (purple). This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 32 Figure 26: The perpendicularity for each ruler is plotted as its angle’s error from 90°. Vendors A – J are displayed from top to bottom, with vendor A being at the top (yellow) and vendor J at the bottom (purple) Bending Upon receipt of the new rulers, we noticed that many of them had varying degrees of bending (Figure 27). Figure 27: Rulers had varying degrees of bending out of the packaging. The DMG studied the effect of bending on the length scale. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 35 Figure 30: Measured height difference due to bending over the center-to-center distance between circles. The bending varies significantly among vendors. To remove confusion, we plot each cluster of data points separately (Figure 31). Please note that each graph has different scales for both the x- and y-axes. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 36 Figure 31: Clustered data points for height difference are isolated to provide more detail. It should be noted that the plots present varying x- and y- scales. We notice that the largest bending occurs at nearly 2mm height difference over an 84.9 mm center-to-center length segment. To determine the impact of bending to the length scale, we calculate for each ruler. Since the bending is approximated to be linear and the length scale is located along the center-to-center length segment, the angle between graduations is also . So, to calculate the corrected length scale distance between zero and any subsequent centimeter graduation, we will use the following trigonometric relation: where is the length between zero and the graduation in question as measured by the vision machine, is the corrected length for that segment, and is the angle of bending. The geometrical representation of this trigonometric relation is given in Figure 32. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 37 Figure 32: The corrected distance between zero and the x graduation, , can be determined using the above trigonometric relation, given the measured distance, , and the angle of bending, . Since is an even function (symmetric about ), negative values of have the same effect on length scale correction as positive values of . This allows us to work with absolute values of . Below we present the largest measured bending for each vendor (Table 6). Since the correction is linear along the scale of the ruler, we present the correction to the length scale as a value per centimeter. We also present the correction at the 5 cm graduation as it will be the largest correction. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 40 “Maintenance tolerances in excess and in deficiency for measures except metal tapes shall be as shown in Table 1: Maintenance Tolerances, in Excess and in Deficiency, for Linear Measures Except Metal Tapes. Acceptance tolerances shall be one-half the maintenance tolerances.” Table 1. Maintenance Tolerances, in Excess and in Deficiency, for Linear Measures Except Metal Tapes Nominal Interval from Zero Tolerance feet inch ½ or less 1/64 1 1/32 2 1/16 3 3/32 4 1/8 5 5/32 6 3/16 Table 7: “Table 1” from NIST Handbook 44, Section 5.52, Part T.1 provides minimum accuracy requirements for length measurements by rule or tape. The first row is highlighted as it covers the length range of length scales in this study. Note that acceptance tolerances are given to be one half the maintenance tolerances. It should be noted that, in many standards and specifications, including NIST Handbook 44, it is common to have two tolerances: acceptance and maintenance. The maintenance tolerance is larger and reflects the general assumption that rulers can wear, which can cause slight length changes. This is most often caused by damage to the scale marking lines. We look to the tolerance for rulers with a nominal length of ½ feet or less, which is 1/64 inch. Since this value reflects the maintenance tolerance, we will take half of it to be the acceptance tolerance, or 1/128 inch. We compare this tolerance with the ABFO No. 2 tolerance in Table 8 below. The values are presented as a percentage of nominal length. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 41 Acceptance Tolerances NIST Handbook 44 and ABFO No. 2 Specification Acceptance Tolerance (%) NIST Handbook 44 0.13 ABFO No. 2 1.00 Table 8: The ABFO No. 2 acceptance tolerance is compared to that from Handbook 44. Tolerance values are presented as percentages of nominal length. We see that the ABFO No. 2 tolerance is considerably higher (by a factor of 7) than that of the ordinary ruler used in ‘legal for trade’ as established by Handbook 44. However, we also note that most of the rulers tested in this study maintained their error from nominal to within 0.2% (0.1 mm) at the 5 centimeter graduation. The worst case exhibited an error within 0.4% (0.2 mm) over the same nominal distance. Another industry standard is one set forth by the International Organization of Legal Metrology. The OIML R 35 – 1 ‘Material measures of length for general use’ is one of several “international recommendations which are model regulations that establish the metrological characteristics required of certain measuring instruments and which specify methods and equipment for checking their conformity.”14 Under Section II - Metrological Requirements, Part 4.2 Maximum permissible error on initial verification, under rated operating conditions: “ 4.2.1 The maximum permissible error on initial verification, positive or negative, a) for the nominal length, and b) for any other distance between any two non-consecutive scale marks, is expressed by the formula: where: L is the value of the length in question, rounded up to the nearest whole 14 OIML (International Organization of Legal Metrology) R-35-1, Material Measures of Length for General Use, Edition 2007 (E) This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 42 number of metres, a and b are coefficients the values of which are given, for each accuracy class, in Table 1.” Accuracy class a b I 0.1 0.1 II 0.3 0.2 III 0.6 0.4 Table 9: “Table 1” from OIML R 35 - 1, Section II, Part 4.2 provides coefficient values for the calculation of maximum permissible error under several accuracy classes. The coefficients are used in the formula , where L is in meters, for any two non- consecutive graduations (or “scale marks”). We examined what the acceptance (“initial verification”) tolerances would be under OIML R 35 - 1 for the 5 centimeter graduation (Table 10). Acceptance Tolerances at 5 cm (OIML R 35 - 1) Accuracy Class Acceptance Tolerance (mm) I 0.2 II 0.5 III 1.0 Table 10: The OIML R 35 - 1 acceptance tolerances are calculated at the 5 cm graduation for each accuracy class. We compared the acceptance tolerances at 5 cm with the ABFO No. 2 tolerance. The values are presented as a percentage of nominal length (Table 11). Acceptance Tolerances at 5 cm OIML R 35 – 1 and ABFO No. 2 Specification Acceptance Tolerance (%) Accuracy I 0.4 Accuracy II 1.0 Accuracy III 2.0 ABFO No. 2 1.0 Table 11: The ABFO No. 2 acceptance tolerance is compared to tolerances at 5 cm for each accuracy class in OIML R 35 -1. Tolerance values are presented as a percentage of nominal length. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 45 This shows that there are commercial alternatives to NIST for calibration, which would have large savings in time and cost. It also shows that the tolerances from a current standard could be used to simplify a standard for forensic photography scales. The bulk of a new standard could be used to better define the materials allowed for scales, criteria for the dimensions and legibility of markings, and optimizing targets for photography. V. Notes Note 1: The ABFO No. 2 technical note also provides specifications for width of legs and reflectance values of grayscale feature. These features were not considered for this study as they were deemed not quantitatively significant. Note 2: The ABFO No. 2 technical note gives two conflicting specifications regarding the external diameter of the circles: In the third page of the technical note The Bite Mark Standard Reference Scale – ABFO No. 2, under the section titled Scale Development, it reads: “Three circles, each 20 mm in diameter, are included on the scale.” The excerpt is shown below as a cut from the digital file. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 46 While, on the fifth page of the technical note, under the section titled Analytical Considerations and Techniques, it reads: “The internal and external diameters of the three circles are 19.75 and 23.0 mm, respectively.” The excerpt is shown below as a cut from the digital file. Note 3: More examples of the variation in print quality among rulers. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Page | 47 Disclaimer: Certain commercial equipment, instruments, or materials are identified in this paper in order to specify the experimental procedure adequately. Such identification is not intended to imply recommendation or endorsement by the National Institute of Standards and Technology, nor is it intended to imply that the materials or equipment identified are necessarily the best available for the purpose. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice.