What is in those compact shelves?

Auditing and quantifying a government documents collection

Jennifer Kirk is government information librarian at Utah State University’s Merrill-Cazier Library, email: jen.kirk@usu.edu

As academic libraries evolve, the complexity and breadth of federal government documents collections pose challenges to librarians. These materials are often shelved under the Superintendent of Documents classification system (SuDoc) and, depending on the library, may be only partially cataloged. The collection’s monolithic nature may complicate projects to catalog, promote, move, or deaccession materials. Using a collection inventory to define mini collections by SuDoc area helps quantify these large collections for academic librarians. The collection inventory system described in this article can be customized and applied beyond the scope of government information. Its approach informs how we assess and build collections, how we inherit collections, and how we work within established systems.

Like many other academic libraries, Utah State University (USU) participates in the Federal Depository Library Program (FDLP). It joined in 1907 alongside other land-grant institutions. Government documents are stored by SuDoc in compact shelving, microfiche cabinets, map drawers, and oversize areas. As a result, the collection is hidden on closed shelves or in drawers. USU government documents collection policies from the mid-1960s indicate that cataloging was not a priority. Staff relied on commercial indexes and shelf lists instead. There has not been a comprehensive approach to cataloging the collection. Current cataloging primarily relies on batch record uploads from Marcive and small-scale projects. In 2017, collection statistics indicated there were approximately 480,000 documents. However, eyeballing the materials indicated this statistic was not accurate. Without fundamental information about the collection, the library was unable to perform basic maintenance, update signage, and conduct minor shifting effectively. In 2018, government information staff undertook an inventory that focused on defining the collection by smaller areas. This inventory resulted in a better understanding of the extent of our collection, the gaps, and the creation of a patron-friendly visual aid explaining the collection.

A collection-based approach

SuDoc classification organizes materials by originating agency. Using SuDoc areas, such as “I” for the Department of Interior, to define smaller collections emphasizes natural divisions within the larger collection. This compartmentalization allows librarians to identify needed description (cataloging or a shelf list) for discrete areas. It also provides faster, simpler description of the extent (measurement of materials), which archives use to provide a basic understanding of the amount of materials. Because government documents collections exist across academic libraries, extents can be compared to determine if an area contains a lot or a little of an agency’s publications. Mini collections based on SuDoc allow librarians to examine what a specific agency may have published and prioritize projects to build or deaccession materials accordingly. Tools like the Guide to U.S. Publications1 and an agency list prepared for the HathiTrust Registry of U.S. Federal Government Documents2 are essential for providing context for agency-produced documents at the SuDoc level.

The government information collection in compact shelving at Utah State Univeristy Libraries.

The government information collection in compact shelving at Utah State Univeristy Libraries.

Baseline data and preparation

Inventories are guided by what library workers need to know about their materials.Inventorying a partially cataloged government documents collection requires additional consideration. Intensive inventories that map catalog records to physical materials3 or generate shelf lists from catalogs4 are not useful for a partially cataloged collection. Technological approaches using barcode scanning or RFID are nonstarters for historic collections that lack those components.5 Literature on preparing collections for moves offers helpful information6 but does not aid in collection development considerations for a holistic approach to this work. An inventory at USU Libraries needed to provide baseline data that would address these challenges while allowing for collection development.

The inventory prioritized shelf composition and measurement of materials (extent) over a physical count. As it was designed to collect information about shelf contents and identify preservation needs, the inventory categorized types of shelves: mostly “Paper” items, mostly “Bound” items, a “Mix” of the two, and shelves containing 3-Ring Binders. These shelf-composition counts were later used to calculate estimated counts as described in the results section below. Using this method, the collected shelf data is separate from the count averages, and results can be refined with additional information as needed.


USU Libraries used Google Forms for our inventory, and a copy of the survey is openly available.7 Using an online form for the inventory allowed for controlled data entry and structured how info-gatherers moved through the stacks. Pilot testing proved that it was easy to update the form. Finally, we could copy the survey and export the data without affecting data collection.

To prepare for the inventory, we needed adequate staff time and materials. Patron usage decreases in the summer, making it the opportune time to conduct the project. An iPad mini and a smartphone were used for data collection. Info gatherers used numbered Post-its placed on range ends to track their progress, initialing when complete. This told us where we had been and who completed the work if there were follow-up questions.

The survey instrument included four major areas:

  • Shelving: row number reviewed and total number of shelves.
  • Collection management: shelf characteristics, shelf capacity, and signage.
  • “Count” questions: used to derive extent calculations and estimate quantities of materials.
  • Inventory management: confirm completeness and form logic.

Question formatting ensured data collection would be relatively quick, consistent, and reviewable. Whenever possible, the form used multiple-choice or drop-down options. Yes/no questions allowed for follow-up when necessary. This sped up data collection while controlling data entry. The info gatherer would start at the beginning of a row, identifying both the row and the SuDoc area, then progress to the end of the SuDoc area. They would submit a new form for every SuDoc area in a row. Some rows would only have one submission, while rows with multiple SuDocs would have up to 12. Images helped data gathers answer questions that were more complex. Each shelf type included an example image in the form. Finally, info gatherers confirmed completeness within the instrument, ensuring proper communication and embedding a quality control mechanism.

Training and pilot testing were critical. Info gatherers were experienced student staff who knew the inventory’s goals and anticipated outcomes. They participated in form design during the project’s pilot test. As a result of feedback, we moved some questions, clarified wording, and added a question about shelf capacity. Training also imposed limits to time spent reviewing the shelves. Info gatherers were to review for only two hours per day. Without such constraints, the form had a gamification quality to it, incentivizing review and potentially compromising accuracy. Time limits also ensured that other work, namely processing incoming materials, continued at nearly its regular pace.

Following initial data collection, a round of data verification examined potential inconsistencies. Sometimes, info gatherers clicked the wrong SuDoc or entered inaccurate numbers. These entries required resubmission. Submissions needed to total the number of shelves counted and fit within our expected number of shelves per side (66-to-70 shelves). Prior to the inventory, we assumed an error rate of about 10%. We needed to verify about 7% of entries. Because responses were segmented by SuDoc area, we were able to quickly verify problems and obtain accurate data. The inventory took approximately three weeks.


As designed, the inventory produced counts of shelf contents and extent. The results were easily combined into collection profiles based on SuDoc area. From that data, we estimated the quantity of materials in our collections from each originating department. The estimated number is derived from a formula that multiplied count of shelf type by the average number of materials for each shelf type. We generated custom averages from six SuDoc areas (A, E, I, LC, S, and Y.4) to produce a rough estimate of materials. The shelf type amounts were totaled to obtain the estimated number of materials per SuDoc area. Accuracy was a challenge with very small SuDocs, so we did not use baseline averages for areas that were a shelf or less. Instead, the inventory identified the location, and we obtained manual counts for those SuDocs. The estimated count assumptions that we applied are not embedded in the form or survey itself, so the survey form could be adapted for review within larger SuDoc areas. For example, in Interior (I) or Congress (Y), the inventory methods could further divide these large SuDoc areas into smaller stems based on originating agency or committee (e.g., I 19).

The inventory has helped quantify the large government documents collection into mini collections for colleagues, administrators, and patrons. For example, our Department of Agriculture collection (A), which supports many stakeholders at our land-grant institution, is 506.5 shelves. Our average shelf length is 2.875 feet, so that is about 1,456 feet of materials. We estimate that we have about 67,000 items in our print Agriculture collection. In contrast, our Department of Labor collection is 140 shelves, 403 feet, or approximately 17,500 items. The inventory produced such data for every SuDoc area in our collection. It also identified SuDoc areas that are not represented.

This inventory produced actionable results that we can build upon. It identified maintenance needs: updating signage, fixing sliding materials on shelves, and replacing three-ring binders. We can focus on smaller areas and properly document projects. Additionally, we now know enough about specific areas to employ techniques used elsewhere. The University of Mississippi employed a sampling technique to identify misplaced items on their compact shelves,8 and the University of New Mexico reviewed individual items over a period of years.9 This collection-level inventory prepared us for similar item-level review.

This inventory’s key takeaways are adaptability and efficiency. In segmenting our collection by SuDoc area, we have allowed for re-use and resurveying. Adjustments after pilot testing ensured that we did not encounter a problem after days of work. In the future, we could use the same techniques to identify item-level preservation needs. This inventory process addressed limitations imposed by our compact shelving. We could not have two people in the same range of shelves at the same time. Open stacks would allow for the methods described here to be deployed simultaneously in different areas with proper inventory tracking. Ultimately, we developed a method for reviewing a large collection efficiently.


The Government Information Department used data from this inventory to create an infographic highlighting and contextualizing the government collection for patrons.10 It gives approximate collection totals and displays our extent in terms that are easy for visitors to understand. It lists our highest collected agencies, and it tells a little bit about FDLP. The Government Information Collection, once hidden in the stacks, is now easily visible for tour leaders and fellow librarians. Now, we focus on the collection instead of the moving shelves.

Most importantly, we have a better understanding of the collection and its individual parts. This inventory was useful in identifying where action is needed, determining where to start, and justifying funding requests to library administration. It has helped to better articulate what is in the stacks and has introduced an efficient way to quantify monolithic government documents collections in academic libraries across the United States, even if they are only partially cataloged.


  1. Donna Batten, Guide to U.S. Government Publications, 2017th ed., 3 vols. (Farmington Hills, Michigan: Gale, 2017).
  2. “Master_agency_list,” Google Docs, accessed February 19, 2020, https://docs.google.com/spreadsheets/d/1EsrPBRT4u3lKkbvdt7X4au4Jf9MmN-xB3nrWox-ILxs/edit#gid=1.
  3. Barbara Braxton, “Inventory: 26 Reasons for Doing One,” Teacher Librarian 32, no. 4 (April 2005): 52–53.
  4. Martha Fallahay Loesch, “Inventory Redux: A Twenty-First Century Adaptation,” Technical Services Quarterly 28, no. 3 (May 19, 2011): 301–11, https://doi.org/10.1080/07317131.2011.571636.
  5. Judith M. Nixon, “The Right Book on the Right Shelf: Missing and Misshelved Books—How Bar Code Scanning Inventories Can Solve the Patrons’ Dilemma,” Collection Management 34, no. 4 (October 1, 2009): 276–84, https://doi.org/10.1080/01462670903175221.
  6. Elizabeth Chamberlain Habich, Moving Library Collections A Management Handbook (Westport, CT: Greenwood Press, 1998).
  7. “Gov Info Shelf Inventory,” Google Forms, accessed February 19, 2020, https://goo.gl/forms/8Kfus3EM1tviPOfG2.
  8. Judy T. Greenwood, “Taking It to the Stacks: An Inventory Project at the University of Mississippi Libraries,” Journal of Access Services 10, no. 2 (April 1, 2013): 77–89, https://doi.org/10.1080/15367967.2013.762266.
  9. Laura Kohl, Claire-Lise Bénaud, and Sever Bordeianu, “Finding Shelf Space in an Academic Library: A Multifaceted Approach,” Technical Services Quarterly 34, no. 3 (July 3, 2017): 268–82, https://doi.org/10.1080/07317131.2017.1321378.
  10. “Collection Infographic,” USU Libraries, accessed February 19, 2020, http://libguides.usu.edu/ld.php?content_id=52454401.
Copyright Jennifer Kirk

Article Views (Last 12 Months)

No data available

Contact ACRL for article usage statistics from 2010-April 2017.

Article Views (By Year/Month)

January: 12
February: 14
March: 5
April: 7
May: 5
June: 11
July: 7
August: 2
January: 21
February: 12
March: 15
April: 15
May: 9
June: 5
July: 13
August: 10
September: 20
October: 15
November: 17
December: 4
January: 0
February: 0
March: 0
April: 0
May: 0
June: 0
July: 0
August: 0
September: 0
October: 0
November: 329
December: 47