More historians, researchers and even general students of the war reach for The Official Records of Union and Confederate Armies than any other set of published records. And for good reason. No where else will you find in one collection more reports, correspondence, casualty lists, and other important primary material than the Official Records (or as it is more commonly referred to, the OR). Access to this mammoth set of 128 individual volumes plus the index has to date posed a significant problem for many researchers. Just finding a complete set is not always an easy task since relatively few individuals own it. In addition, not all libraries possess (or display) them, in large part because the thirty-some feet of shelf space required for display often makes owning a set impractical. The computer age has solved several of these concerns, for now the entire set of OR is being offered for sale on CD-Rom by three separate publishers at essentially the same time. While not everyone owns a computer, and fewer still own one with CD-Rom capability, the number of these machines available is growing. I must admit I enjoy holding and browsing through the books themselves. There are few things finer in life than cracking open a new book and enjoying the wonderful scent of a freshly printed page. Thus, the thought of using a computer to dig into a reference set that I have sitting on a nearby shelf was not initially appealing. Being a lover of the physical product I have not, by and large, changed my mind, although the ability to pop in a disk and conduct a thorough search offers significant advantages and is also an enjoyable process.

The publishers that have submitted their product for review are: Broadfoot Publishing Co. of Wilmington, NC; H-Bar Enterprises, of Oakman, AL; and Guild Press, of Indianapolis, IN. Both Broadfoot and Guild stressed that their products were advanced prototypes and subject to some changes, while H-Bar's was a finished piece of software. While all three versions were found to be professionally packaged and handsomely presented, they varied in their ease of usefulness, availability of special or enhanced features, and accuracy.

All three software packages were clear and easy to read on the screen and scrolled smoothly through the material. All three possess point and click icons as well as drop-down menus, which make it easy for the user to access the available options and resources offered by the software. Since it is impossible in the limited space available to fully review every feature of all three packages, I have decided to offer a brief discussion as to how each package functioned generally before focusing on a few major features that most users will find of interest in determining whether to purchase this reference set on disk, and if so, which to acquire. I examined all three packages within a Windows95 environment using a Pentium 133mhz machine and 32 mb of ram. No obvious system conflicts were observed.

The first package I examined in-depth was the prototype offered by Guild Press, which contained the majority of the OR volumes. Guild's gray and blue screen is divided into two portions (left and right), with the viewing screen on the right, and an index tree/root system on the left. The respective screens can be easily sized to the convenience of the user with the mouse, and various search windows may be opened at the same time. [Guild] The OR are listed in the left window by serial number, although the standard method of citing this set is by the actual volume number and not the serial number. I wanted to access vol. 27, pt. 2, pg. 18 (Gettysburg reports), for example, and experienced some minor delays until I figured out that serial 44 was the volume I needed. By using the point and click method, I chose this serial and was rewarded with a second window below the first listing 100-page sections and the reports contained in this volume, arranged by number. Each section contained a "page jump table" spanning 100 pages ("topic breaks," as the publisher calls them). Clicking on this table produced a calendar-like table with each page denoted. Clicking on the page sent me where I wanted to go. The Guild prototype allows for "simple" text searches (which seeks out exactly what you typed), and "complex" searches (which employs searches using "near", "or", "and" and "not.") Other searches can be launched within certain fields, such as the signer of reports, the addressee, or the date of the writing. My simple search for "Dan Tyler" resulted in numerous search possibilities listed in the left lower window by serial volume (in this case No. 44, for example), with the "hits" showing up on the right side of the screen highlighted in black. Clicking on an arrow button on the top strip menu sent me forward or backward to each strike, although I could not determine how many "Dan Tyler" strikes had been found by the Guild search engine. Guild's prototype also offers the user a breakdown by "Great Battles" and "Significant Court Martials/Inquiries."

I next examined Broadfoot's prototype, which contained only the three Gettysburg volumes. This package, also with a primarily blue and gray interface, contained one large screen with a top button strip and drop down menu options. [Broadfoot] Searches are initiated by clicking on a flashlight icon, which pops up a box with all the commands on a single screen, including search operators (and, or, not, near, within, and a pair of parentheses), a comprehensive indexed word wheel for use as a search guide, and a current field guide (main, text, or header). By typing in "Dan Tyler", the search engine kicked out a two line message that read: "82, Serial No. 44 Series 1 Vol. 27, Part II reports." which means that there are 82 separate references to "Dan Tyler" in vol. 27, pt. 2. The second "hit" line evidenced an additional 56 strikes in vol. 27, pt. 3. Clicking on the first line took me to the title page of vol. 27, pt. 2, while a click on either of a pair of yellow arrow buttons took me forward or backward through the listings, with "Dan Tyler" highlighted in yellow. The individual page breaks were in blue text and easy to see.

The last of the three I examined was H-Bar's product, which also contained the three Gettysburg volumes. H-Bar's software, also with a blue and gray interface, is designed with the point and click buttons arranged along the left side of the screen and drop down menus across the top. Clicking on the query button popped up a screen with a box within which to type the search text, an indexed search wheel, and an "infobase" scope guide, which lists numerous search fields similar to Guild's. Typing "Dan Tyler" brought up a page of text on the screen with the words highlighted in blue. At the bottom of the page were three small ribbons listing the record number "14322/59404," the number of hits as "1/66" (meaning the first of sixty-six, although Broadfoot's had discovered 138) and "Query: dan tyler." Cycling through by clicking on a pair of black arrow buttons took me to the Tyler strikes. Individual report headings are reproduced in bright purple.

While all three versions are based on allowing users to search using key words, Broadfoot's is also equipped with what it calls "hypertext," a function that allows the user to search for words highlighted with a mouse. This proved to be a very helpful attribute. Guild's provides a similar feature by allowing users to highlight words, copy to clipboard, and paste into the query field. All three packages allow for proximity searches ("Hood and Lee within 25 words" for example), although it appears that such searches in Guild's prototype are conducted only within individual reports or correspondence and not across the entire OR.

After operating all three for some time, it became apparent that all three packages operated with similar principles and methods, similar interfaces, and without significant differences. This is not to say that the packages are indistinguishable, for they are surely not. Since the OR is usually referenced for research purposes as opposed to pool side reading - and therefore cited in scholarly studies - the ease of citation will be a major concern to users. When this set is only visible on a computer screen, and page by page at that, is it simple to determine which volume, part (several volumes are broken into two or more "parts," or separate books), and page number you are viewing? This threshold inquiry revealed the first substantive difference in the three products.

While both Broadfoot and H-Bar clearly denote which page a reader is on, only Broadfoot's has the complete citation (series, serial number, volume, part, and page), at the top of every page in bright blue lettering, which is easy to see. H-Bar reproduced the headers printed at the top of each original OR page, but these citations provide only the page number and chapter - but not the volume or part, critical information you need in order to properly cite from the OR. As it turned out, the information as to volume and part was available in H-Bar's version. I had to scroll up and down several pages to find it, however, because it is printed, like the text itself, in black, which makes it hard to locate. Similarly, Guild lists the volume and part number in a static window at the top of each page, but this window does not include the page number the user is on. Instead, the page number, substantially smaller than the surrounding text, is embedded in the text of the page itself and surrounded by brackets thus: [657]. This rather cumbersome method of pagination forces the user to comb the page for a very small page number that I found difficult to locate. No one wants to spend precious research time just trying to figure out which page (or volume and part) you are searching.

The precision of reproduction from book to computer disk is another aspect I spent considerable time examining. I selected several pages at random from each software package and compared them to the printed page. Let's take vol. 27, pt. 2, p. 18 as an example. Two of the publishers - Broadfoot and Guild - impressed me with their completeness and accuracy. I could not find a mistake in either's version. H-Bar's package, however, had the most errors in transcription (be it electronic or manual). While some were of the niggling variety, several were of a more serious nature. For example, on the page cited above, there were a half-dozen of what I would consider minor but troubling errors (extra spaces in front of commas, "3,30 o'clock" instead of "3.30 o'clock"; the letter "m" standing alone as a separate word in the middle of the text where none exists in the original version; and the word "the" dropped from the middle of a sentence). Substantive inaccuracies, obviously more serious than those heretofore mentioned, were also present in sufficient number to raise significant reservations. For example, H-Bar's version reproduces Thomas A. Maulsby's name as "Maulby" several times on page 18, although it is spelled correctly on page 19. Similarly, the One Hundred and Twenty-Sixth Ohio (126th Ohio) is noted as "Once hundred and twenty-sixth Ohio"; "throw" is spelled "thrown," and "on our front with at least a brigade..." is reproduced as "on out front witch al least a brigade". [H-Bar] These types of errors denote hasty and sloppy editing. While several pages of H-Bar's software I examined did not have any errors (as one would hopefully expect), the number of them found on this single page alone would force me, as a researcher, to double-check substantive citations in the original set of books. This step, of course, defeats most of the purpose of owning the software in the first place. If you can't confidently depend on the software to reproduce the original OR, can it be used for research and citation purposes? It is also interesting to note that, (as near as I could tell), paragraph breaks in both the Broadfoot and Guild versions mirror exactly those found in the printed OR, while H-Bar's does not. Thus, the five separate paragraphs that appear on vol. 27, pt. 2, page 18 are reproduced in H-Bar's package as one long paragraph. This is important if someone wants to cite a long passage in quotation.

The transcription and presentation of information formatted into tables, such as casualties and strengths, also varies considerably between these three publishers. The importance and usefulness of the data found in tabular form, which speckles the OR landscape only somewhat less frequently than meteor craters on a barren lunar body, cannot be overstated. Thus, the manner of its presentation on the computer screen is of no little importance. I randomly selected the single margin-to-margin table found in vol. 27, pt. 1, p. 294 as an example. Both Broadfoot and Guild reproduce it, like the original, as a single table, while H-Bar breaks it into two separate tables, one for officers, and the other for enlisted men. Of the three, Broadfoot's method of displaying tabular data is clearly superior to its competitors in view of its ease of understanding and printing. It appears to have been retyped and re-formatted (as was Guild's) using a standard legend (k=killed, KO=killed officers, etc.), so that it would fit, like the original, as a single table. The major difference between Guild's and Broadfoot's is that Broadfoot's table easily printed out onto a single page of paper, and printed exactly as it appeared on the screen. Guild's version ran off the right margin (albeit slightly). Although Guild's version is slightly cramped and not as attractive or instantly intelligible as Broadfoot's, both are accurately reproduced with regard to the information contained thereon. H-Bar's version printed fine, but the two-table formatting system makes it much more difficult to understand. Of more importance is the disappointing fact that the figures under the columns denoting casualties for the officers and enlisted men were not properly lined up. Thus, an unsuspecting user might run afoul of some Pennsylvania Gettysburg buff when claiming that Paul's First Brigade lost 35 officers killed at Gettysburg on July 1, when in fact the original table claims but a single such casualty. This type of formatting error is a serious problem because it renders the table worthless. Lastly, the original table contained a brief footnote denoted by an asterisk (*). All three publishers reprinted the footnote text (and reprint footnote and explanatory items found throughout the OR). Both Broadfoot's and H-Bar's appear, as in the original version, on the page itself. Guild employs pop-up text boxes that are accessed via asterisks (*) with the mouse button.

While they are often not particularly good, the original OR has dozens of maps that are useful when thumbing through the reports and correspondence. It is natural to expect that these maps would be available in these computer versions of the OR. Only two of the three publishers, Broadfoot and Guild, reproduced them. H-Bar's package does not offer any maps at all, a strange oversight. Of the two versions available for comparison, the differences were relatively minor but worth noting. Take, for example, the full-page map found in vol. 27, pt. 1, p. 915. Guild chose to print the map (apparently the same size as the original) sandwiched between two blocks of text, thus it will only print with the surrounding text. While this is not a serious flaw, it is not an accurate representation of the original version. Broadfoot's presentation, while more faithful to the original, is not perfect either. Although printed as an entire page, it is not clear what page, for citation purposes, the map is on. Broadfoot's programmer promises that this oversight on the prototype will be corrected for the final version. Both publishers thoughtfully enhanced the original maps by allowing a user to zoom in and out on each map by clicking on the right button on the mouse, which makes it easier to read some of the fine print. Guild's used gray scale scans, which makes it easier to read small text on enlarged maps, while Broadfoot's monochromed scans appeared a bit jagged in this regard. Broadfoot's has two additional features designed to allow users better access the OR maps. The first is a skipping mechanism so a user can jump back and forth from map to map (without intervening text) by simply clicking on a camera icon found at the top of the page. The second is the added feature of allowing the user to rotate the image in 90 degree increments. These last pair of features make Broadfoot's offering in this regard better than its counterpart, although overall Guild's presentation is virtually identical.

Attempting to actually print selected on-screen material also developed significant differences between the three software packages. All three have simple point and click printing functions, but selective printing was only possible with two of the three packages (Broadfoot's and H-Bar's). In other words, a user can simply highlight a single word, sentence, paragraph or page and print that selection. With Guild's prototype, however, there is no option to print selected text within the OR software package. A user thus has two choices: print the entire topic selection (several pages long), or highlight a slice of text, paste it onto a separate word processing program, and then print the selection from that piece of software. Given the obvious time, effort and expense that Guild has spent developing its prototype, this extra step is an irksome (and unusual) formatting quirk.

While all three programs allow users to easily copy text and paste it into another document (Word Perfect or Microsoft Word, for example), Broadfoot's package includes what it calls an "electronic notepad," a separate window-based system which allows users to copy, paste, add their own "marginalia," or notes, and save for later use (or print into a hard copy version), all without having to rely or use another word processing program. Guild's includes a "bookmark" feature that permits a hierarchical bookmark list to be made. This data is stored in a single file which can be exported and printed with another application. I could not find any way to do anything similar with H-Bar's product.

While I have tried to focus on several key areas that will hopefully interest potential users and purchasers of these important products, there are other custom features that deserve mention. The most important of these are offered by Broadfoot. According to their computer representative, they entered "thousands" (their number) of corrections and additions, all of which are presented in italicized text throughout the OR. These corrections include such things as spelling errors, errors in rank, and so on, while the additions might offer a middle name where only an initial (or nothing) was available in the past. Also, if a user wants to jump to a specific page in a single step, as far as I can determine, only Broadfoot's search engine allows you to do so by specifying the volume and page number. While all three versions are based on allowing users to search using key words, Broadfoot's is also equipped with what they call "hypertext," a function that allows the user to search for words highlighted with a mouse.

The cost of this software is not inexpensive. Guild's has the lowest retail cost at $450*, which includes Dyer's three-volume Compendium and Fox's Regimental Losses, two very helpful and welcomed sources on CD-ROM. H-Bar's price for the full set was the highest at $750*, although you can purchase individual battles (1-5 books, depending on the battle, on one disk), for about $50.00 each. These are placed on a recordable CD and can be mixed with several other offerings, including Jefferson Davis:Constitutionalist, Rise and Fall of the Confederate Government, to name just two. Broadfoot's full set of the OR is $600.00. Broadfoot also claims it will have several Civil War scholars "scrutinize" the first edition of its software for mistakes, and that it [Broadfoot's] will send, free of charge, an updated version to everyone who purchases this first edition. While I have not seen it, Broadfoot will offer a complete printed and bound user's manual with its finished product. Guild will apparently offer a scaled down version and rely more on its on-line assistance, while H-Bar does not offer any printed instructions. All three software packages offer on-line help, and all three offer a money back guarantee.

Is it worth purchasing the OR on CD-Rom? After several hours of experimentation with all three packages, the advantage of sitting in one chair and searching the entire OR with a global search engine became increasingly obvious. Printing hard copies with the click of a mouse (as opposed to lugging heavy books to a copier and plugging the thing with nickels), coupled with the ability to send valuable information into your favorite word processing format (i.e. pasting text directly into your manuscript) are just two major reasons why all serious researchers will have to have a full set of the OR on CD-Rom. Although it takes getting used to, digging into the OR on disk will eventually prove faster, easier and cheaper than using the printed version. I hesitate to say that using the mammoth 128 volumes is now outdated, since it is a question of taste and long years of habit. The real battlefield of this war between these three publishers is a question over the custom features each offers (or doesn't offer) and the accuracy of translation. Each potential purchaser will have to decide for himself which options are more important for his own particular purposes.

FAST AND INTUITIVE RETRIEVAL

Retrieve text and images (pictures, maps, sketches, etc.) quickly.

SIMPLE ONE-WORD OR PHRASE SEARCHING TO MULTI-WORD BOOLEAN SEARCHING

Our search engine allows a search of one word (i.e., "Lee") or full-name and multi-word phrases (i.e., "Robert E. Lee" or "The Greatest Battle of the World"). Boolean searching expands or narrows the search field using linking words like AND, OR, NOT, NEAR, WITHIN (i.e., search for "Lee and Grant" or search for "Gettysburg" appearing near the word "Lee," etc.).

RELEVANCE RANKING

Search "targets" are listed by the frequency of occurrence or "hits" allowing you to concentrate on the most productive text.

WORD WHEEL

An alphabetical list of all the indexed words in the text. The Word Wheel lets you easily determine which words are in the text (rather than searching) and suggests words which are similar to your search word. This is helpful if you're not sure of the exact spelling or differences in spelling which have occurred due to archaic word use or misspelled words; especially names and place names.

HYPERLINK

Electronic links in the textbase which enable you to easily and rapidly move forward or backward in the textbase without having to employ cumbersome "scrolling" or "page up" and "page down."

HYPERTEXT

Makes it easy to select any word in the text as your search word.

PROXIMITY SEARCHING

Search for words appearing in the text within any specific number of words from each other; be it 5 or 999 words (i.e., find where "Lee" and "Jackson" occur within 12 words of each other [approximately 1 line] or within 999 words of each other [approximately 1 page].

IMAGE RETRIEVAL

Images such as maps or sketches can be searched, retrieved, displayed, and printed. All maps and sketches have been fully annotated, indexed and hyperlinked so you can easily find them.

TEXT AND IMAGES CAN BE PRINTED OR DOWNLOADED

Text (up to 999 lines) and images can be downloaded to a personal file, a printer for printing, or a disk for future use.

SEARCH HISTORY

As you use our program, a listing of the searches you've made is kept (up to 32 searches from the most recent to the oldest). This feature allows you to quickly search for a word or phrase you've previously searched for without re-keying the search specification. These search histories may also be "saved" for future use.

TABLE OF CONTENTS SEARCHING

Identifies individual volumes by content allowing you to access individual volumes and scroll or search them as if you were using hard copy (books).

FIELD SEARCHING

Each page consists of three fields; header, text and main. You can search any or all of these fields at one time.

---

• Is Page for page identical to the book version
• Contains all 128 Volumes - 131,000 pages
• Provides completely visible citations