Bryn Mawr Classical Review 95.09.27


R. Biering and V. Brinkmann, Dyabola. Elektronische Sachkataloge zu den Altertums- und Kunstwissenschaften. Ennepetal: Biering und Brinkmann, 1994. 1 CD-Rom & 1 programmdiskette.


Reviewed by John Tamm, McMaster University (g8110765@mcmail.cis.mcmaster.ca).

Searching through bibliographies can easily become a time-consuming and otherwise frustrating affair; any aid to such searches is therefore highly desirable. Developed by R. Biering and V. Brinkmann, the menu-driven software package Dyabola -- the DYnamically Accumulating dataBase on Objects and Literature about Antiquity -- is such an aid, and is therefore much to be welcomed. As this is still a fairly new and unfamiliar -- in North America at any rate -- program, the discussion that follows is both an explanation as well as a critique of the program's operations.

At its heart lie the bibliographies of the DAI in Rome (from 1956 onwards, dealing with antiquity in general), the Römisch-Germanische Kommission of the DAI in Frankfurt (generally from the mid-1980's onwards) and the DAI in Madrid (from 1990 onwards, specialising in Iberian archaeology). Each of these constitutes a separate database, whose data can be accessed through the various program functions. The user should be aware that as these are archaeological bibliographies, works deemed to be outside the scope of the project, such as those dealing purely with philology, are not included.

There are several possible ways to search these databases: by subject, keyword, author, title, series, periodical or part of a title. Information can also be added to the database, and then in some cases be used as part of the search criteria. Search results are stored in temporary user-named datafiles which can then be displayed on the screen and/or sent to the printer and/or saved permanently for later recall. In setting the operating parameters, the user can specify whether these results are to be arranged in chronological or alphabetical (author surname) order. Searching in chronological order is the faster of the two, but the difference is not significant unless the search is based on a criterion so broad as to render the resulting datafile virtually useless due to its size. It is also possible to perform some manipulations on these datafiles. Two can be combined using one of a number of logical operators and the result, stored in a third user-named file, can then be displayed, printed, saved or even manipulated further. Finally, search results can be exported from Dyabola, in the form of an ASCII text file.

The screen display is not limited to just the basic bibliographical information about the entry being displayed. The user will also be notified of the number of possible search paths for that entry under the 'Subject' category and the number of reviews of that entry. Pressing the appropriate key will cause the actual paths or the reviews to be displayed. Moreover, if the entry refers to a collection, for instance a Festschrift or the proceedings of a conference, the user will be notified of this as well, and will have the chance to view immediately the other articles in the collection.

The default language of the program is German, while English, French and Italian are also supported. One can change the language used from within Dyabola, but only for the duration of that particular session; a permanent change of the default language is also possible, as will be shown later. Some knowledge of German -- of the basic archaeological terms at any rate -- is nevertheless helpful, as not everything is translated into the language selected by the user. This becomes especially noticeable when one runs searches in the 'Subject' category -- the abbreviations for the search terms remain in German, although the full forms are given in the language selected by the user, and the 'Keyword' category - terms which are not names of people or places remain in German. Greek is also supported, in so far as one is able to enter search terms, for instance titles or author names, in Greek.

As can be expected, such a program is not inexpensive. Our package cost a total of 4,122.50 DM. For this we received a 3.5" diskette containing the programs needed to install and run Dyabola, a CD containing the datafiles -- editions 1-4 with the supplements for 1991, 1992 and 1993 -- and a user's manual, concise, helpful, but a few years old and therefore not entirely up-to-date.

Installation:

As a minimum platform, the makers demand -- in their manual -- an 80286 running MS-DOS 4.0, with 640K of RAM and 150 MB available on the hard disk (should one wish to install all three databases). The monitor needs a graphic standard EGA, VGA or Super-VGA card. The only printers supported are those from the Epson LQ series (or those which are fully compatible with the Epson LQ's). A mouse, unfortunately, is not supported.

The minimum system, however, would probably be too slow for most users; it would definitely by now be inadequate in terms of memory -- the databank for the Rome catalogue alone takes up almost 152 MB, while the package as a whole, programs and datafiles, takes up almost 180 MB. Moreover, the program needs temporary space to store the search results, and the user may want to save some of these results, further adding to the memory requirements. A CD-ROM drive is also now necessary, since as of spring 1994 the datafiles are available only on a CD.

At McMaster we are using a Dell Dimension XPS P90, a Pentium-based system with an internal speed of 90MHz, running MS-DOS 6.21. It has 8 MB of RAM, 256 KB of cache memory and a 512 MB hard disk (drive C). We have no worries therefore about memory limitations, and search speeds are amazingly fast. For disk drives, we have one 3.5" drive (drive A) and one CD-ROM drive (drive D). The monitor (colour) is a Dell UltraScan 17ES with an SVGA card. An Epson LQ 510 serves as a printer, while an HP LaserJet 4MP is available for use if the data is exported from Dyabola into, for instance, a wordprocessing program.

Installation is relatively simple and quick; on our system, the process took about 20 minutes. A looseleaf insert with both English and German text gives the required procedures, superseding those in the manual. One first copies the contents of the 3.5" diskette onto the hard disk, using

xcopy a:\*.* c:\ /s /e < Enter >
This copies all the directories and files needed by Dyabola, save the data files. One could save disk space and retrieve data directly off the CD, but in the interests of speed, the makers recommend that the data files also be copied onto the hard disk. To do this, one must first switch to the appropriate directory, and then copy the files. To copy the Rome catalogue, for instance, one enters:
cd \dya\dyabola\dya_rom\ex < Enter >
copy D:\dya\dyabola\dya_rom\ex < Enter >
For the Frankfurt and Madrid catalogues, one substitutes 'dya_rgk' or 'dya_mad' for 'dya_rom' as appropriate. There are some inconsistencies in the insert: on the English side, the initial '\' is left out of the 'cd' command, rendering it invalid, while the German side adds the destination drive, 'C:', after the 'copy' command, which is not strictly necessary.

The directory tree is not very complicated; one should however mention the four separate directories named 'dyaerg'. The first is on the path 'c:\dya\dyaerg', and contains the subdirectory 'tmp_####', where # represents a digit, in which Dyabola temporarily stores the results of the searches. Normally, these datafiles will be erased on program exit; should the program crash during its run, however, -- as has happened to us on a few occasions -- this will not occur, and instead, more subdirectories named 'tmp_####' or 'tmp_###' appear. Should this happen, these additional subdirectories with their files should be deleted through DOS, to avoid cluttering up the hard disk. The other directories named 'dyaerg' lie on the paths 'c:\dya\dyabola\dya_rom (or dya_rgk or dya_mad) \dyaerg' and contain the search results saved by the user from within Dyabola. These are permanent datafiles; those that are no longer needed ought again be deleted through DOS.

The batch file D.BAT controls the program run. A few modifications to this batch file may be helpful. As mentioned earlier, one can change the language used by the program from within Dyabola, but only for the duration of the session. If one wants to change the default language permanently, one should edit the line in D.BAT reading:

dya /L1 /P3=3D1 /B changing L1 to L2 (for English), L3 (for French) or L4 (for Italian). The /L1 switch may be missing, as in our program; in this case, one simply inserts the appropriate switch. The second modification concerns the way Dyabola terminates. It always leaves one in c:\dya, regardless of the directory from which it was invoked. The addition of a 'cd' statement at the end of D.BAT may therefore be desired. In our case, 'cd \' gets us back to the root directory on drive C.

Program operation:

Dyabola is invoked by pressing < d >< Enter >. This brings up an introductory screen; pressing any key will bring up the first menu, consisting of the names of the three databases as well as the 'quit' command. Before selecting the database to be searched however, the user may want to press < Shift + F10 >, to access the parameter setup menu. From this menu the user can change the program language for the duration of the session, initialise the printer (not really necessary as this will be done automatically the first time output is sent to the printer), change the sort order from chronological (the default setting) to alphabetical order, enable polyhierarchic retrieval -- to be explained later -- and enable a feature called 'global keyword index.' This last option remains a mystery -- it is discussed neither in the manual nor in the on-line help, which exists for many of the program screens and can be called up by pressing < F1 >.

As with almost all the menus in Dyabola, an option is selected by moving the cursor, a bar which highlights the option it rests over, to the desired selection with the up or down arrow keys, and pressing < Enter >. It is unfortunate that only one parameter can be changed at a time; once a selection has been made, the user is returned to the database selection menu. To return to the setup menu, < Shift + F10 > must then be pressed twice.

Although the setup menu can be accessed from anywhere in Dyabola, the sort order option is available only when this menu is accessed from the database selection menu. The user must be aware that accessing the database selection menu will have the same effect as exiting the program -- all search results which have not been saved as permanent datafiles will be deleted.

Once a database has been selected from the database menu, a title screen for that database is displayed. Pressing any key will bring up the main menu screen. In the upper left-hand side of this screen is a box containing the options, while down the right-hand side is a box that will display the names of files resulting from searches, the loading of previously saved files and the combination of such files. Beside each filename will be a number specifying the number of entries in that file. Across the bottom of the screen is a bar containing the names and functions of the currently enabled keys. In general, there will be a similar bar across the bottom of all the program screens. The provision of help in such a way is a nice feature.

Before moving on to a discussion of the individual program operations, a few general comments should be made. First, the program is case and accent sensitive. When prompted for input therefore, the user must enter capital letters and accents where appropriate. A table showing how to form accents is included in the manual.

Next, for each search the user will be prompted to enter a name for the resulting datafile. Despite the instruction on the screen that only letters are accepted, numbers can be used as well. The name is limited by the program to a maximum of eight letters, and duplicate names are not allowed. The same comments hold true for the names of files that the user saves permanently.

For each search, the user will also be prompted to enter a range of dates in which the search is to be carried out. If < Enter > is pressed with no specific range entered, the search covers the whole timespan of the database. If a single year is desired, one can enter either the full four-digit or the shortened two-digit form -- both '1990' and '90' are for example acceptable. For a range one must specify both start and end dates, even if the end-date is the last year included in the database (currently 1993), for example: '1990-93' or '90-93'.

Error checking of the dates is not as comprehensive as it could be. Although 1956 is the earliest stated date for entries, the program will accept a date as early as 1951. At the other end, dates from 1994 up to 1999 are also accepted even though the databases stop with 1993. This is not too much of a problem if a range of dates has been entered, provided that part of the range falls in a valid timespan -- program operations are not hindered; if a single invalid year is entered though, the program will conduct the search all the same.

Another annoyance is that the dialogue box into which the user must enter the date range can appear anywhere on the screen. Program operations are not affected, nor does it matter at this point if some of the information on the screen is obscured, but a single, consistent location for this box would be preferable.

Finally, one should mention that the user will at times be asked to select a datafile or name or term from a list of such items. The selection is made just as a menu item is selected -- the cursor is positioned over the desired entry, and < Enter > is pressed to make the selection. This list may be lengthy; in such a case, the < PgUp > and < PgDn > keys can be used to speed up the selection process. These move the cursor up or down one window's worth of entries at a time.

In the examples to follow, the DAI Rome database has been used as the source database unless otherwise noted. While names and terms may change for the other two databases, the basic principles will remain the same.

i) Main menu -- searches by subject

A search under this option will produce all the literature available on a given topic for the chosen date range; a great deal of detail is available, so that the search can be made very specific. Choosing 'Subject' will first bring up a list of general topics, for example 'Mosaics' or 'Vessels'. Topics appear in two forms: an abbreviation in German, and a full name in the language set for the session. This list extends over two screens, and the user can scroll through it in the usual way. Neither at this or any subsequent level of topics is the user told that the topic list extends over more than one screen, even though this happens occasionally; it would be nice if such information were displayed.

A final selection could be made at this stage by pressing < Shift + Enter >, but this is not advisable -- there would be too much data. The 'Vessels' topic, for example, contains 13,293 entries in all. As an aside, making such a choice does demonstrate the speed of the program. A search in chronological order under 'Vessels' took only 6 1/2 minutes, while a search in alphabetical order took 8 minutes.

It is therefore necessary to make the search as specific as possible. The user will notice an asterisk just to the left of the abbreviation for all but one of the topics. For this level as for all the subsequent levels, an asterisk if present means that further subdivision of the topic and therefore greater specificity of the search criterion is possible. To access these more detailed levels, < Enter > is pressed. Once the topic has been narrowed down as much as is desired or possible, < Shift + Enter > is pressed to set the search path. If there is no asterisk beside a topic, then that level is as detailed as possible for that topic. One can, of course, begin the search at any level by pressing < Shift + Enter >.

To continue with the 'Vessels' example, the next level includes such topics as: 'terracotta', 'glass', 'metal', 'shapes', and so on. Many of these are the last level of specialisation; some, such as 'terracotta', lead to further sub-topics. 'Terracotta', for instance, leads to 'Orientalising', 'black-figure', 'red-figure', 'coarseware', and so on. Again, many of these are already the most detailed level while others have yet more sub-topics. 'Black-figure' for example leads to the categories 'Attic', 'Lakonian', 'panathenaics', and so on. With this list, the end of this branch of the subject tree is finally reached.

Some topics will have a pair of square brackets following the term. The search path '*iconographic themes - > *mythology - > single myths or figures [ ]' can be taken as an example of this. In such a case, when the topic with the square brackets has been selected, a window with a list of further terms will appear, and a selection must be made from this list. These terms are what the program calls 'keywords', about which more will be said later. In this case, the selection would define the myth or figure desired. Once this stage of the search path has been reached, it is no longer possible to abort the search. Contrary to the instructions in the manual, the user will not first be asked for input; the window will open with the cursor positioned over the first entry in the list and the user must scroll through it until the desired term is reached.

At first glance, it would appear that there are two possible paths should one wish to use a specific site as a search criterion. This is not actually the case. The path '*topography - > *sites [ ]' is a little confusing. If < Enter > is pressed at '*sites [ ]', the user will be able to select one of Athens, Istanbul or Rome. These sites, I believe, are what this path is meant to be used for. Should < Shift + Enter > be pressed at '*sites [ ]', a window with a very long list of individual sites will indeed appear, from which a selection must be made. However, this path does not seem to have been included in the subject headings for the data entries; at any rate, the results (if any) obtained by using it will not be satisfactory. The correct path to follow is '*topography - > *countries'. This will bring up a list of countries, from which the appropriate country can be selected. For some countries, no further subdivisions are possible. For others, one can choose between 'regions [ ]' or 'sites [ ]' to end up with a much more specific search term.

Besides the method outlined above, there is also another way to access a specific search term, using the < F2 > key. Once pressed, the user will be prompted for the desired term (or part), which must be entered in its German form. A small window with a list of search terms will then appear; beside each term in this list is a number, which indicates the number of times that particular term occurs in the subject heading tree. Once a selection has been made from this list, the user will be taken directly to the menu that contains the chosen term. The display will be the same as if the method described previously had been followed, so the full search path can be seen at a glance. If the term occurs more than once, pressing < Shift + F2 > will take the user to the next occurrence, until all the occurrences have been displayed. At that point, either a selection must be made from the menu then displayed, or < Esc > must be pressed to return the user to the main subject headings list. There is no provision for going back through the occurrences.

A nice feature is that the windows for each level of topics do not overwrite each other, but instead cascade down the screen, so that the topic for each window is displayed. Thus, the user can easily tell where he or she is in the subject tree, and what the path taken to get there was. Also useful is the fact that the user can back up through the levels with the < Esc > key, should a wrong path have been followed.

Truly distressing is the fact that the program has on three separate occasions crashed during searches under this search option. As has been mentioned, < Shift + Enter > is used to make a selection, while < Enter > opens up the next level of topics. If < Enter > is pressed when a topic has no further sub-topics, the cursor generally returns to the first entry in that particular topic list. This is what is meant to happen. On one occasion, however, when defining a search path under the 'Vessels' category, rather than returning to the top of the list when < Enter > was pressed inadvertently (at '*vessels - > *terracotta vases - > geometric vases') the screen instead went blank and then the DOS prompt appeared. Upon re-entering Dyabola, following the same search path, and deliberately pressing < Enter > in other inappropriate locations in this menu, the same result occurred. Other search paths were unaffected. In another session about a week later, again when < Enter > was pressed at the wrong time, a similar crash occurred. This time, the search was being conducted under the 'Mosaics' topic. Why the program should have crashed at these times, when it otherwise handles this situation correctly, remains a mystery. Even more mysterious was the third crash, which could not be duplicated. In this case, the search path was '*epigraphy - > *Greek - > *acta publica'. Here, < Enter > was pressed correctly, at the term '*acta publica'. Instead of presenting the next level of topics, again the program terminated and one was returned to DOS. Three crashes, three different search paths, and no obvious explanations. I can not be sure about the second crash, but it will be seen that both the first and third crashes occurred in the second level of subject headings down from the initial subject heading list. There clearly seems to be some sort of bug at this point in the program, which occasionally makes its presence known. The fact that such a crash could happen again is worrying. It must be said though that a loss of unsaved search results and a proliferation of directories named 'tmp_####' or 'tmp_###' in c:\dya\dyaerg seem to be the only side effects of these crashes. Program operations and the integrity of the datafiles have not been compromised. Moreover, this crash has occurred only in searches of the DAI Rome database.

At this point, a slight digression is in order. Dyabola is not a Windows application, but provided that a suitable PIF file is developed for Dyabola, it can be run from Windows just like a Windows application is. Having installed Dyabola on Windows, I have found that the abovementioned crash can be guaranteed if one presses < Enter > at any term, in any menu, that is the second level of menu down from the main subject heading list. If < Shift + Enter > is pressed instead in this level of menus, thus making a selection, the program works normally. As before, this crash only occurs in searches of the DAI Rome database.

It would be appropriate also to return to the subject of polyhierarchic retrieval. As explained in the manual, identical subject headings may appear along different branches of the subject tree. With this feature enabled, all the entries for a particular heading, no matter where in the tree the heading is, will be retrieved. This seems very much like a search under the keyword category, but the two are not identical, as will be shown in the next section.

Finally, I should point out that for the DAI Madrid database, only German abbreviations have been used in the subject headings - the full English equivalents are missing.

ii) Main menu - searches by keyword

Keywords differ from subject headings in that proper names, of figures from antiquity and of distinguished modern scholars, as well as conference venues and the classical terms for art- objects and items from daily life are included in addition to place names and names of mythological figures, such as can be accessed through a subject heading path. Depending on the search requirements, it may sometimes be more useful to search under a keyword than a subject path.

As the manual points out, the keyword 'Delphoi' for example can be reached, in a search under the 'Subject' category, by more than one path: '*topography - > *countries - > *Greece - > sites [ ]' and '*religion - > myths and cults of single countries and places [ ]'. There may even be more possible paths. Thus, all must be searched if an exhaustive list of the works concerning Delphi is desired. With a search under the 'Keyword' category, the user need only specify 'Delphoi', and all the paths will be searched automatically.

Having selected 'Keyword' from the main menu, the user will be prompted to enter the desired keyword or part thereof. A window containing the list of keywords will then appear, with the cursor positioned on the keyword that is the closest to the user's input. The correct keyword must then be selected.

As mentioned earlier, searching under this category is almost, but not quite equivalent to searching under the 'Subject' category with polyhierarchic retrieval enabled. A search under 'Keyword' will retrieve more entries, sometimes significantly more. Using 'Delphoi' as a test case, for the date range 1980-90 a search under this category retrieved 130 entries, while a search under the 'Subject' category retrieved 110. Similarly, a search under the keyword 'Herakles' produced 51 entries for the date range 1990-93, while only 21 were retrieved in a search under the 'Subject' category. This is understandable in that a work touching on Delphi or Herakles need not be found only in a search path terminating in 'Delphoi' or 'Herakles'.

iii) Main menu -- searches by author

To search for a particular author's works, the 'Author' option must be selected. The user will then be prompted for the author's name. Either the whole or just the beginning of the surname may be entered; an initial or initials separated by periods may also be included if the whole surname is entered. Upon pressing < Enter >, a window with a list of names beginning with the letters entered by the user appears, from which the user selects the desired name. The initial letter of the surname must be capitalised, or else the user is sent to the end of the list of names beginning with that letter. All accents must also be included.

This search finds not only all the works by an author, but also all the reviews he or she has written (within the parameters of the database). This explains why some of the results found may at first glance appear to be irrelevant -- they are works which the author has reviewed rather than written.

A few criticisms can be made about the operation of this search option. There seems to have been no consolidation of the various forms with which authors have signed their works. Thus, to get a complete list of the works of J. D. Beazley or K. M. D. Dunbabin, for example, one must search under both Beazley, J. and Beazley, J. D., and all three of Dunbabin, K., Dunbabin, K. M. and Dunbabin, K. M. D. Similarly, one should check both accented and unaccented forms of a name, just to be sure. The works of Y. Thébert, for instance, are to be found under both Thebert, Y. and Thébert, Y. Moreover, I am quite sure that Keppie, L., Keppie, L.F.J. and Keppie, L.J.F. refer to the same author. Many more such examples can be found, if one browses through the author list. I accept that in some cases it may be difficult to determine whether two or more signatures belong to the same author, but surely in most cases such matters could be resolved.

Another slight problem arises from the fact that the three databases must be searched independently. Thus, it may be necessary to search all three of them to get a complete list of an author's works.

iv) Main menu -- searches by title

Searching for a specific title is also possible. After choosing this search option, the user will be asked to input the desired title (or part). Besides the usual need for following the capitalisation and accentuation of the original title, a leading article, if present, must also be included. Having entered this data, a window with a list of titles will appear, with the cursor positioned over the title closest to the one inputted by the user. The cursor should then be positioned over the correct title if necessary, and the selection made. A number will appear beside each entry in the list. This indicates the number of works with the same title.

A little dialogue box asking whether or not to include all subdivisions of the database in the search will appear after the title has been selected. This question seems superfluous here, as a specific title rather than a general subject heading is being used as the search criterion. At any rate, answering either 'Y' or 'N' seems to have no effect on the performance of the search.

v) Main menu -- searches by series

If a work is part of a series, such as the supplements to the JHS for example, it can also be found through this search option. When selected, this option causes a window containing a list of the names of all the series included in the database to be displayed. The user is not asked for input before this list is displayed, unlike the other searches that first require a selection from a list. This is a bit bizarre; the user is always placed at the beginning of this (long) list, and is forced to scroll through it until the desired title is found.

vi) Main menu -- searches by periodicals

Choosing this option will produce a list of the contents of a selected periodical over the date range specified. The user will be asked to indicate whether an abbreviation or the full form of the name will be used. Upon input of the abbreviation or name (or part thereof), a list of either abbreviations or names of journals will be displayed, from which the desired journal may be selected. In general, a periodical will be found either under its abbreviation or under its full name. The AJA however can be found under both, so there are some inconsistencies.

As mentioned earlier, articles deemed to be outside the scope of the program will not be listed. In what I trust is an isolated case, an article that should have been listed in the contents of the JRS for 1973, the review by Frederiksen of Vermaseren's work on the Mithraeum at S. Maria Capua Vetere, has been left out of the database altogether.

vii) Main menu -- searches by any part of title

Even if only a part, any part, of a title is known, a search can still be made. The user will first be asked to enter the string to be used as the search criterion. Then, a sub-menu will appear, asking whether the string to be sought for should match the user input exactly, or whether capitals or capitals and accents should be ignored. The user is also given the chance to abort the search at this time. A third input will then be required: whether the search is to proceed with or without user confirmation of the entries found. The search then takes one of two paths.

If a search with confirmation was requested, then the entries are displayed on the screen one by one and the user has the chance to save or discard entries as desired. To keep the entry on the screen in the result file, < Shift + Enter > must be pressed. Pressing < Enter > alone causes the entry on the screen to be discarded. Either way, after making the selection the next entry will be displayed. The only disadvantage to this procedure is that the user is not told of the total number of entries that have strings matching the search string; the process could therefore seem endless. However, once the desired title is found, and if there is no desire to browse through the rest of the entries, the process can be stopped by pressing < Esc >. Indeed, pressing < Esc > will abort the search at any time.

If no confirmation was requested, then the search is carried out as usual although the matching entries do flash past upon the screen, and the whole result file is added to the results box. In this case, the progress of the search can be followed because a small bar graph showing the percentage of the database searched will be displayed. If matches are few and far between this works well; if not, the graph flickers too much to be really useful. It is impossible to abort this type of search once started, unless one reboots the computer; inputting as specific a string as possible is therefore recommended.

viii) Main menu -- searches by local data

This option enables searches based on call numbers, publishers, 'corporate bodies' and ISBN numbers, through the 'Local Data' sub-menu. Selecting one of the first three categories again causes a window with a list of possibilities to be displayed, from which the appropriate selection must be made. For the ISBN category, one must enter the desired number (or part) oneself, and a final choice must be made from the list then displayed.

This whole option has very much the look of a 'work-in-progress.' While a good number of publishers have been included, very few of the database entries actually have a publisher listed. The results of a search based on a publisher will not, therefore, be very comprehensive. For example, a search based on the OUP produces only three entries for the whole DAI Rome database. Similarly, only two 'corporate bodies' are listed, both Spanish, while a mere handful of ISBN numbers are included, in whole or part.

No on-line help is provided for this option, while only the call number category is mentioned in the manual. This is unsatisfactory. What, for instance, is a 'corporate body', and why is it worth distinguishing in this way? When would a search based on ISBN numbers be desired? It seems hard to believe that no other information would be known for a work.

Much better is the handling of this option in the DAI Madrid and DAI Frankfurt databases. ISBN numbers have often been included, in full, with an entry, and the list of ISBN numbers to search for is much fuller. Publishers, too, have been included more frequently in the entries. Finally, there is a comprehensive list of 'corporate bodies' to search under; here one can see what is meant by this term -- the DAI Berlin is one, for example, and the J. Paul Getty Museum, Malibu another.

ix) Main menu -- displaying search results

To view the results of a search, the '-- display result' option must be selected. The cursor will then move over to the results box, from which the appropriate result file can be selected. A large window will then open up in the upper left-hand side of the screen, in which the entries from the selected file will be displayed in the order, alphabetical or chronological, initially chosen by the user. The entries can be scrolled through forwards or backwards, using the < PgDn > and < PgUp > keys respectively. The end of the file presents no problems; pressing the appropriate key twice will restart the process.

There is a bit of a bug in the display process, apparent only if the entries have been arranged in alphabetical order. If one should start viewing them in reverse order, then after the first entry in the file one will see the expected blank window, but next, instead of the last entry in the file, an entry from somewhere in the middle of the file but with the number of the last entry. Should one then start scrolling forwards, one will notice that the entry-numbers continue to increase. Only once the end of the file has been reached and the scrolling process is started anew will the correct sequence and numbering be restored.

Besides the bibliographic information, there will be a number in the upper right-hand corner of the window signifying the number of the entry in the result file. Across the bottom of the window will be a bar specifying the number of possible search paths under the 'Subject' category as well as the number of reviews that exist for the entry. There may also be an asterisk in this bar. If the asterisk is present, it means that the entry is the main reference for a collection. Articles from a collection should be indicated by either an SB (for collection) or KB (for conference proceedings) underneath the main information. Lastly, there may also be an ISBN number, or part thereof, in the lower right-hand corner of the display window.

Before discussing these additional pieces of information, the facility for expanding abbreviations should be mentioned. If an entry originates in a periodical, it is highly likely that the abbreviation for the periodical, and not its full name, will be given in the bibliographic information. To view the full name, the user should press < S > or < s >. A box containing the name will then appear on the screen. Pressing < Esc > or < S >/< s > a second time will return the user to the main display screen. This is a very useful feature; the user should be warned however that this box may appear anywhere on the screen, and in so doing may obscure the information already present. This problem can easily be remedied, by pressing < S > or < s > again. Sooner or later, the box will appear in an otherwise clear area of the screen. This same problem will also be seen when the abbreviations for search paths are expanded into their full forms, and when abbreviations for the periodicals cited in the 'reviews' box are expanded.

There will generally be at least one search path (under the 'Subject' category) for each entry. The path (or paths) can be viewed by pressing < B > or < b >. A small window containing the list of paths will then appear in the bottom left-hand corner of the screen, with the cursor placed over the first path in the list. The window can hold a maximum of six paths; if there are more, the user can scroll down the list in the usual way. These paths are given using the abbreviations seen before, in the subject tree. To call up the full form of a path, the cursor is first positioned over the desired path, and then < S > or < s > is pressed. An unexpected bonus is that in this case, the full path will be given in the user-selected language. Pressing the < Esc > key will return the user to the main display screen.

To display any reviews that exist for the entry, either < R > or < r > must be pressed. A list of the reviews will appear in a window in the bottom right-hand corner of the screen. The information given includes the source and date of the review, as well as the name of its author. As with the search paths, the cursor will be placed over the first item in this list; abbreviations (of periodical titles) can again be expanded into their full forms by positioning the cursor over the desired review and pressing < S > or < s >. The reviews are laid out one per line, and sometimes, there will be too much information to fit on the line. In this case, pressing < S > or < s > will first call up the full information; pressing < S > or < s > a second time will expand the abbreviation. The 'reviews' window can display up to 11 reviews at a time; should there be more, the user can scroll down the list as usual. Pressing < Esc > will return the user to the main display screen.

If the entry is the main reference for a collection, as indicated by the asterisk in the bar running across the bottom of the display window, then the user can call up a listing of the articles in the work by pressing < EMPH > or < EMPH >. The user should also be able to save this list of contents as a separate datafile, for later examination. If the entry comes from a collection, as indicated by SB or KB, then pressing < S > or < s > will call up the main reference for the work.

Our experience has been somewhat different. First of all, while the main references are certainly indicated by the asterisk, entries that should be marked by SB or KB are not. The only way to tell if an entry is an article in such a work is to study the bibliographic information for clues -- such as the title of the work, or a range of page numbers.

Next, pressing < EMPH > or < EMPH > only works for the main entry, not for the individual articles. Thus, for an article suspected to come from a collected work, the user must first press < S > or < s > to see if this is indeed the case. Then, once the main reference is displayed, the user should press < Enter >, which will save the main reference in the datafile the user is in the process of examining. Only then will it be possible to see the rest of the contents by pressing < EMPH > or < EMPH >.

Moreover, the user will not be asked to enter a filename for this table of contents, and the entries in it will not then be saved as a separate datafile. Instead, the cursor will be placed on the first entry in the list. Any articles the user wishes to save must be selected in the usual way, one at a time. The chosen article will then be saved in the datafile the user is examining. The way this is done may seem a little confusing at first, mainly because each selection is not only saved into the file in its proper place, but is then promptly displayed as well. If alphabetical order has been selected as the sort order, the user might end up far from where he or she was in the file. At least one can return quickly to the main reference, by pressing < S > or < s > and < Enter > again. This will not cause duplicate entries of the main reference to be inserted into the file. The insertions may cause the entry numbers to become inaccurate; it is even possible to see negative numbers, if one scrolls backwards through the file. This situation is corrected though once the user has scrolled through the entire file forwards. Finally, the effects of these insertions will not be seen in the results box until the display process has been exited. Then, the number of entries will change to the correct number.

This option is handled a bit better in the Frankfurt and Madrid databases. There at least the signposts SB and KB are included, as well as one more, FS, for an article in a Festschrift.

A potentially very useful feature is the ability to add information, known in Dyabola as 'Local Data' to the entries in the databases. Two classes of information are catered for: first, the call number of the work (or source of the work) in the user's library, and second, whatever additional information the user may wish to add, up to a maximum of 200 characters. To prevent misuse of this feature, it has been password-protected. The user will have to enter a three-character password, included with the program documentation, before being allowed to enter any data, or view any data but the call numbers.

It should be made clear that the actual entries in the databases are not being altered. Instead, local data is stored in a separate file named 'sign_1.btr', which can be found in the 'dya_rom, dya_rgk and dya_mad subdirectories, from which it can be recalled. It is important to make backup copies of these files, especially before installing an update to Dyabola. This updating process will delete the contents of these files.

Rather unfortunately, our package did not include the password. The following paragraphs are therefore necessarily based mainly on the information in the manual which, as I have often had cause to point out, is not totally up-to-date.

In order to enter local data, a search must first have been performed, and the results displayed. When the user finds an entry for which the addition of such information is desired, he or she must first press < Shift + L >, and then enter the password. At this point, a window will appear, which is meant to hold the call number. Pressing < Shift + Enter > confirms the entry and closes this window. Then, another window opens, into which the rest of the information can be entered. Again, pressing < Shift + Enter > confirms entry of the data and closes the window. The information in both of these fields can be edited, by re-entering the fields and deleting or overtyping the data. Deletion of all the local data for an entry is also possible; to do so, the user must enter < - > in the call number field.

Local data can be viewed when the entries of a file are being viewed. There are two options. The call number is accessible to all users; for works which are not articles in a periodical or collection, the call number will be automatically displayed in the upper left-hand corner of the display window, while for works that are, it will be displayed along with the expanded form of title abbreviations when < S > or < s > is pressed. The other information is accessible only to those who know the password, and can be viewed by pressing < L > or < l >, and then entering the password.

The output of result files will be discussed later; it is also possible to output just an entry. There are two ways to do this. The < Print Screen > key could be used, but this would also print out all the other, extraneous, information on the screen. The user should instead press

or

. This will print out just the basic bibliographic information, as seen on the screen.

To delete the entry currently being displayed from the datafile, the < - > or < _ > key must be pressed. The entry number will decrease by one, but the number in the results box will not change until the display process is exited. This is the only direct way provided by Dyabola to delete entries in a file (a side-effect of the 'Combine' operation may be a file with fewer entries, but one doesn't have the precise control over deletion as one does here).

x) Main menu -- Tools

Selecting the 'Tools' option calls up the 'Tools' sub-menu, which makes possible for the user the manipulation, permanent storage and output of the search results. It should be noted that the first time this option is selected in a session, the first item of the 'Tools' sub-menu is highlighted, while the last item selected is highlighted subsequently. This can save some time, but it can also lead to inadvertently selecting the wrong item. The < Esc > key will however return the user to the 'Tools' sub-menu.

x.i) Tools menu -- combine

The 'Combine' option enables the combination of two datafiles in a variety of ways. After selecting combine, the user then selects the first data file to be manipulated. Pressing any key will bring up the sub-menu containing a list of logical operators. After selecting one of these, the second file to be manipulated is specified. The user is next asked to enter the name of the result file. The program will then perform the operation, the result file will be added to the results box and the user will be returned to the main menu.

The logical operators that may be used are 'AND', 'OR', 'XOR', 'DIFF' or 'NOT'. Choosing 'AND' leads to the entries common to both files being written to the result file. This is handy in that if one wants to see all the works of, for instance, author 'X' on subject 'Y', one first searches for all the works of 'X' and all the entries for 'Y', and then combines the two using 'AND'. With 'OR', the two files are simply joined together. Should there be entries common to both files, only one of the pair will be written to the result file, the other will be suppressed. 'XOR' also leads to the files being joined, but common entries are eliminated entirely. To determine the entries in the first file that are not in the second, 'DIFF' must be used. The last choice, 'NOT', is completely useless as it writes to the result file all the entries in the database which are not in the file specified in the operation. If the result would be a blank file, as would happen if, for instance, author 'X' wrote nothing on subject 'Y' or all the entries of the first file were also in the second one, then the user will simply be returned to the 'Tools' sub-menu and nothing will be added to the results window.

x.ii) Tools menu -- save/load/copy/rename/delete

For permanent storage of a datafile, the 'Save' option is selected. A prompt asks for a filename; this can be the same as the search result name, or different. The file to be saved is then selected from the list of search results. To retrieve a saved file, the 'Load' option is selected. Again, the user is first prompted for a name for the file. Then, the list of saved files is displayed and the user makes a selection. This retrieves the file into the temporary file created by the user.

The 'Save' option can use memory rather extravagantly. The program always reserves the same amount of memory, 28,300 bytes, for the file whether it contains one entry or, as extreme a case as I have tried, 13,293 (all the entries under the general subject heading 'Vessels'). It is to be hoped that this situation will be remedied in the future.

Moreover, while the amount of files that can be saved at one time seems to be unlimited, when one uses the 'Load' command only the names of the first 20 such files (in the order in which they were saved) will be displayed on the screen, a point not made in the manual. A message stating that more files exist does appear on the screen, but there is no way to get the names of these files displayed; they therefore remain inaccessible until one deletes some of the saved files through DOS. Then the names of the first 20 of the remaining files can be seen, and so on. This particular error message having once been activated, it now appears when the maximum number of search results has been reached as well -- this should not happen, as the message is inappropriate here.

The 'Copy' command enables the copying of a search result into a newly-created and named file, while the 'Rename' command enables the renaming of a search result.

To delete search results, the 'Delete' command is used. The program will only support 20 search results at any one time, and once the limit is reached, the user will be warned that no new searches can be made. Again, this is not mentioned in the manual. Previous results will therefore need to be deleted, to make room for new ones. One first selects the file to be deleted in the usual way. The user will then be asked to confirm the deletion. Only one file can be deleted at a time. Unfortunately, this command only works on the temporary files. There is no way to delete any permanently saved files within Dyabola; they must be deleted from DOS instead.

x.iii) Tools menu -- print/print lists

These two options provide the major mechanisms for producing output from within Dyabola. The latter works well; the title of the file is printed and below it, the information in familiar list format. The former, however, does not altogether work in the way the makers of Dyabola seem to have planned.

Essentially, what seems to have been the aim is the printing of all the relevant information for an entry -- bibliographic information, the paths by which the entry could have been reached if searched for under 'Subject', and the reviews (if any) of the entry. The bibliographic information appears as it does on the screen. Below it appears the list of possible search paths. Then, the main search path is repeated as a heading, but where the list of reviews should appear there is only blank space. The size of this space does change -- it is directly proportional to the number of reviews of the entry. This leads me to conclude that output of the reviews was planned, but that a bug in the program prevents this output from being realised.

This shortcoming is unfortunate. As it stands, the only way to print out the reviews for an entry is, while viewing the entry on the monitor and having called up the reviews, to press the < Print Screen > key. The disadvantages to this are that all the extraneous material on the screen will also be printed out, but not the search paths, and that only one entry can be printed out at one time. Furthermore, as only 11 reviews can appear on the screen at once, one may need to repeat the process to get all the reviews.

A few other points concerning the printing process also deserve mention. First, while printing is occurring, no other work in Dyabola is possible. Secondly, if one is using single sheets in the printer rather than continuous feed paper, and if one is feeding these sheets into the printer manually, then there will clearly be occasional delays in the printing process as the printer is taken off-line, one sheet is removed and the next one inserted. This does not create any problems; printing will carry on once the printer is put back on-line. After some time, however, the printer will stop and an information box reading 'printer timeout' will appear on the screen. Pressing any key will cause the printing operation to continue. Lastly, the user can tell that the printing process is nearly over when the main menu screen reappears.

x.iv) Tools menu -- export

Search results and files saved in Dyabola are stored in a fashion that is comprehensible only to Dyabola. Should a user wish to append such files to another document, or alter or print them in ways not supported by Dyabola, they must first be exported from Dyabola as an ASCII text file, using the 'Export' command. The file to be exported is selected in the usual way; if it is a previously saved file, it must first be reloaded using the 'Load' command. The exported data is stored in the file c:\DYA\DYABOLA.TXT, from where it can be retrieved by, for instance, a word-processing program.

This is a welcome and needed feature, but it suffers from some serious flaws. First of all, the data is stored in only the one file -- there is no choice, no chance for the user to specify a filename. Once data has been exported, the user would be well-advised to exit Dyabola, and at the very least copy the contents of DYABOLA.TXT to another file because the next time 'Export' is used, the contents of DYABOLA.TXT will be overwritten, with no warning. Equally annoying is the fact that only 100 data items can be exported at one time. The amount of time and effort needed to export a file of a few hundred entries -- not an unreasonable number -- for example, can be imagined, especially given the means of deleting data entries from a file within Dyabola. In such a case, it would be much better if the user could devise a way to split up the original search, so as to give a series of smaller files with preferably no more than 100 entries each. The most obvious way to do this would be by splitting up the overall date range into smaller segments, if possible. This would at least save having to delete excess entries one by one, over and over again.

xi) Main menu -- quit

Selecting quit will take the user back to the database selection menu. From there, one can either select a new database to search or exit the program.

Conclusions:

A good number of criticisms have been made in this review, but it does seem fair to say that Dyabola's developers, at times, have not been able to see the trees for the forest. The bugs should be fixed; they are undesirable, even if they are annoying rather than damaging. Dialogue and information boxes should not dance about the screen, but should be assigned a consistent, set, place where existing screen information is not obscured. The easy export of more than a hundred results should be possible. Memory should be managed effectively. Finally, a laser printer should be supported; it is safe to say that such printers are now the standard. Once these matters have been dealt with, the lesser deficiencies of the program should be addressed.

The manual is another area of concern. What is there is in the main useful, but it is not complete, and is at times outdated. A user has the right to expect that all of the program's features will be explained. If printing up a new manual with every edition of the program is not possible, then at least inserts detailing the changes should be provided, as was done for the installation changes.

These, and other matters -- such as the missing password -- I shall communicate to the program's developers. It should therefore be possible, some time in the future, to make known their responses and, hopefully, solutions to the problems.

Despite these criticisms, the question that really must be answered is whether or not the program does what it has been designed to do. The answer is that, even with the flaws -- of which more may yet be lurking in the program, it does, and very quickly at that. A great deal of information is literally seconds away, and retrieval of it is easy. Enough help is provided, both on-line and on-screen, that the user will rarely get stuck provided the program is used properly. I am able therefore both to recommend the program to potential users, and commend the developers on their achievement, with the proviso that improvements continue to be made. The program falls short of perfection, but it is still very good and very useful, and there is every hope that it will get better in future editions.