MARLAP

The Multi-Agency Radiological Laboratory Analytical Protocols (MARLAP) Manual

Overview

The MARLAP manual is a guid­ance docu­ment for radio­chemistry labo­ra­tories and for plan­ners of proj­ects that involve radio­chemi­cal analy­sis. The manual was devel­oped by a multi-agency work­group with repre­sen­ta­tion from the fol­low­ing federal depart­ments and agencies:

• Environmental Protection Agency (EPA)
• Nuclear Regulatory Commission (NRC)
• Department of Energy (DOE)
• Department of Defense (DoD), with rep­re­sen­ta­tives from the Army (and USACE), Navy, and Air Force
• U.S. Geological Survey (USGS)
• Food and Drug Administration (FDA)
• Department of Home­land Security (DHS)
• National Institute of Stand­ards and Technology (NIST)

There was also representation from the states (Kentucky and California).

The MARLAP manual is a large docu­ment, and it is divided con­cep­tually into two parts (although the printed ver­sion actually has three vol­umes). Part I is in­tended primarily for proj­ect plan­ners and deals with data qual­ity ob­jec­tives (DQOs), meas­ure­ment qual­ity ob­jec­tives (MQOs), and ana­lyti­cal issues that proj­ect plan­ners should be aware of. MARLAP Part II is intended mainly for labo­ra­tory per­sonnel and deals with tech­ni­cal issues at the radio­chem­istry labo­ra­tory including chem­istry, nuclear count­ing instru­ments, qual­ity con­trol, and un­cer­tainty evaluation.

History

The work­group met for the first time in Wash­ing­ton, DC, in August 1995. The first meet­ing lasted only about two days, and there were no more meet­ings for several months, partly because of the temporary shut­down of much of the federal bureaucracy in late 1995. The next meet­ing occurred in Mont­gomery, AL, in March 1996. The group clearly under­estimated the effort that would be required to develop the manual. Sub­sequent meet­ings gradually increased in length from two days to five days and became more frequent. Several meet­ings were held at the head­quarters of the Nuclear Regula­tory Com­mis­sion in Rockville, MD. A few meet­ings were held at the EPA’s National Air and Radia­tion Environ­mental Labo­ra­tory in Montgomery, AL, and a few were held at the USGS’s National Water Quality Labo­ra­tory in Denver, CO. Two meet­ings were held at DOE’s Environ­mental Meas­ure­ments Labo­ra­tory in New York City (now DHS’s Environ­mental Meas­ure­ments Labo­ra­tory). Many of the meet­ings were held at the National Institute of Stand­ards and Tech­nology in Gaithersburg, MD.

The manual went through an internal agency review, further revisions, and later a public review and a formal peer review by the Radia­tion Advisory Com­mit­tee of EPA’s Science Advisory Board (SAB).

The manual was finalized in July 2004 and was officially approved by all the par­tici­pating depart­ments and agencies be­tween August and December of 2004.

Training

The first pilot train­ing course for Part I of MARLAP (proj­ect plan­ning) was pre­sented in Mont­gomery on 8 March 2005. Since then several more com­plete 2-day and 3-day courses have been presented.

NOTE: If you plan to attend a MARLAP course, please bring a cal­cu­lator. Some of the course modules involve exer­cises, and some of the exer­cises do require calculations.

Currently no training course is offered for Part II of MARLAP.

How to Obtain the Manual

To view or download the MARLAP manual, visit http://www.epa.gov/radiation/marlap/index.html. It is pos­sible to down­load any or all chap­ters and appendices.

Errata

Listed below are some known errors in the MARLAP manual.

Acronyms and Abbreviations

• In Keith’s opinion the unit symbol for the tropical year (31 556 926 s) should be “a,” not “y”; how­ever, there seems to be no final authority on the sub­ject. The sym­bol “a” is rec­om­mended by ISO 31-1 (Quantities and units — Part 1: Space and time), but a search of the web finds that the same sym­bol is some­times used for other types of years, such as the Gregorian year (365.2425 d). Note that no ver­sion of the year is cur­rently accepted for use with the SI, although some ver­sion is obviously needed.

Chapter 5: Obtaining Laboratory Services

• Page 5-16. Cited References. The most recent revision date for ANSI N42.23 was (and still is) 1996. Accord­ing to a knowl­edge­able source, the docu­ment was re­approved with­out change in 2003; however, the ANSI web site shows the revision date as 1996.

Chapter 6: Selection and Application of an Analytical Method

• Page 6-36. References. The most recent revision date for ANSI N42.23 is 1996. The docu­ment was reapproved without change in 2003.

Chapter 8: Radiochemical Data Verification and Validation

• Pages 8-21 and 8-31. Some of the rec­om­men­da­tions of Chapter 8 are in­con­sis­tent with those of Chapters 1, 16, 19, and 20. In par­ticu­lar, Chapters 1, 16, and 19 allow a lab to report either the com­bined stand­ard un­cer­tainty or an ex­panded un­cer­tainty as long as the un­cer­tainty is clearly explained. (This is con­sis­tent with the GUM.) Chapter 8 rec­om­mends an ex­panded un­cer­tainty only. Further­more, Chapter 8 ex­plicitly rec­om­mends that the lab report a sample-specific MDC with each result. Chapter 20 con­spicu­ously avoids such a rec­om­men­da­tion, noting the like­li­hood that a sample-specific MDC will be misused. The principal author of Chapter 20 intended to ensure that no part of MARLAP ex­plicitly rec­om­mended or forbade the report­ing of sample-specific MDCs. The work­group rec­og­nized that sample-specific MDCs would con­tinue to be reported regard­less of its rec­om­men­da­tion, and chose to em­phasize instead a cor­rect under­stand­ing of the MDA/MDC con­cept.
• Page 8-31. Bibliography. Every other chapter calls this sec­tion “References,” not “Bibliography.”

Chapter 19: Measurement Uncertainty

• Page 19-1. Table of contents. The rec­om­men­da­tions are in Section 19.3.9, as stated in the chap­ter over­view. The refer­ences are listed in Sec­tion 19.6, and there is no Sec­tion 19.7.
• Page 19-24. Equation 19.14 is writ­ten in­cor­rectly. It is cor­rect as writ­ten only when the sen­si­tiv­ity co­effi­cients for each pair of cor­related input esti­mates have the same sign. The prob­lem is that the individual com­po­nents of the stand­ard uncer­tainty are always posi­tive, re­gard­less of the signs of the sen­si­tiv­ity co­effi­cients. This equa­tion was included only to illustrate con­cepts and obviously was not checked care­fully enough before publi­ca­tion. It may be deleted from future editions of the docu­ment, since it is not actually nec­es­sary for cal­cu­lating an un­cer­tainty. Foot­note 8 on page 25 also requires some changes in wording.
• Page 19-30. Example 19.12. The second equa­tion, which cal­cu­lates u_c(p), the com­bined stand­ard un­cer­tainty of p, uses an in­cor­rect value for u(y). The stand­ard un­cer­tainty of y is actually 3, not 4. The rest of the equa­tion, including the final result is cor­rect; so, the error seems to be only a typo.
• Page 19-58. Cited Sources. The most recent revision date for ANSI N42.23 is 1996. The docu­ment was re­approved with­out change in 2003.
• Page 19-99. Apparently the author and the reviewers had never heard of “con­ven­tional mass.” Accord­ing to OIML R 111-1, the values of mass stand­ards are to be given as con­ven­tional mass, not true mass. (ASTM follows suit.) This subtlety can make a slight differ­ence in how one cor­rects for buoy­ancy when the balance is calibrated in terms of con­ven­tional mass. Buoy­ancy effects in general are usually small, and the dif­fer­ence here is really tiny, because mass stand­ards tend to be dense. In the buoy­ancy example, the density was 7.85 g/mL and the refer­ence density for con­ven­tional mass is 8 g/mL. The cor­respond­ing rela­tive error in the buoy­ancy cor­rec­tion fac­tor would be only about −3 × 10⁻⁶. If the true mass of the stand­ard is used for cal­ibra­tion, as the author of Chapter 19 apparently assumed, the buoy­ancy cor­rec­tion fac­tor is OK as is. Never­the­less, the con­cept of con­ven­tional mass should be men­tioned in the section.
• Page 19-100. Equations 19.65 and 19.66. The next revision of the chap­ter should cite the CIPM formula (1981/1991) for the density of moist air (source avail­able in French at http://www.bipm.fr/en/publications/pub_divers.html) and use an approxima­tion more con­sis­tent with it.
• A new definition is needed for a log-normally dis­trib­uted quan­tity on page 19-68. Tech­nically, one can­not take the loga­rithm of the value of a quan­tity un­less that quan­tity is dimen­sion­less. So, the new def­ini­tion might say that a quan­tity X has a log-normal dis­tribu­tion if the logarithm of its numerical value is nor­mally dis­tributed. The “numerical value” of a quantity (with respect to an appro­priate unit of meas­ure­ment), unlike the “value,” is always dimensionless.

Chapter 20: Detection and Quantification Capabilities

• Page 20-34. References. The stand­ard num­ber for the last refer­ence on the page should be ISO 11929-3, not 11929-2. (Thanks to Manuel Martinez Moreno for point­ing out this one.)

Appendix G: Statistical Tables

• There are no known errors in the down­loadable PDF ver­sions of Appendix G, but the printed copies have Arabic charac­ters sub­sti­tuted for Greek let­ters. If you notice similar prob­lems in your PDF ver­sion, please con­tact Keith using the link below.

If you find other errors in the manual, par­ticu­larly in Chapters 19 and 20 and Appendices C, F, and G, please e-mail Keith at Work@McCroan.com. Keith also has particular interest in Chapters 1, 3, 6, 16, and 18; how­ever, he will accept comments on all parts of the manual and pass them along to the workgroup.

Other Likely Improvements for Revision 1

Chapter 6: Selection and Application of an Analytical Method

• Given the results of work per­formed in 2007 and dis­cussed at the RRMC in Jackson, Wyoming, the method valida­tion accept­ance cri­terion is likely to be replaced with a more power­ful sta­tis­ti­cal test based on mean squared error.

Chapter 20: Detection and Quantification Capabilities

• A few of the equa­tions for detec­tion limits may be re­writ­ten in dif­fer­ent but equiva­lent forms.