This page outlines some of the limitations inherent in the data offered by IPNI and summarises the improvements in data quality which have already been achieved. An understanding of the limitations of the data originating from the different sources is essential for wise use of the combined data set.

Understanding the Index Kewensis data

Since the hard copy Index Kewensis was compiled and published over a period of more than one hundred years, significant variations in data entry practices can be detected. These variations, together with information of publication dates, editors and compilers of the index from 1893 - 2000 are described in a table. Click here to view. The information is also summarized as follows:

1. Ranks covered. Until 1970 the only ranks listed are genus and species. From 1971 all ranks from family down to infraspecific level are listed. However, only ranks recognized by the International Code of Botanical Nomenclature are included. There is generally no mention of cultivar names, but some cultivated taxa appear in the early volumes: 'Hort.' - that is hortorum (of gardens), or hortulanorum (of gardeners) - can be a pointer to such taxa. Additionally, genera of hybrid origin, generally horticultural and mainly in Orchidaceae and Cactaceae, are included as long as their publication is in accordance with the International Code of Botanical Nomenclature.

2. The 'Kew Rule' was applied to entries in the original Index and to Supplements 1-III. Under this rule, priority within a genus was reckoned from the date when a specific epithet was first associated with that generic name. Older epithets, previously associated with species placed in other genera, were ignored (Stevens 1991). This practice is not in accordance with the International Code of Botanical Nomenclature. Jackson (1887) described and attempted to justify his use of this rule as follows:

"Whilst on the subject of fixity of names, I would remark that our practice is to take the name under which any plant is placed in its true genus as the name to be kept up, even though the author of it may have ignored the proper rule of retaining the specific name, when transferring it from its old genus to the new, when at least such a name is not already in the genus receiving the accession. To wantonly set aside the joint name thus given, and to publish the new name by joining the oldest specific name to the true generic, is a mischievous practice which should never be condoned; it is adding to the already vast mass of useless synonyms, and is more likely to be the offspring of vanity than of a sincere desire to promote science."

3. Geography. Abbreviated Latin names of countries were used until Supplement XIV. For example: Afr.Lusit.Or = Portuguese East Africa (now Mozambique); Geront.trop = Old World Tropics. Some of the terms are somewhat vague and obscure. Eventually we hope to provide a full glossary here. For the moment, however, the most useful reference work is Stearn (1983).

The geographical annotations in the original two volumes were, according to Jackson (1924), provided by Sir Joseph Hooker.

4. Families. The endings and abbreviations of family names have been standardized and appear in full. Thus Orchideae from the original is now ORCHIDACEAE. However, the given family has not been changed. For example, Elaeocarpus in the original volume and until Supplement III is placed under the family name TILIACEAE, but thereafter it becomes ELEAEOCARPACEAE. The next phase of IPNI will include efforts to assign all names at generic level and below to the families in which they are treated in Brummitt’s Vascular Plant Families and Genera (1992). Eventually we hope to be able to offer alternative views of the data reflecting whichever published system of classification is preferred by the user. In the interim, however, the user is strongly advised to avoid specifying a family whenever possible. For example, if you wish to retrieve all the records for species of Elaeocarpus it should be sufficient to enter Elaeocarpus in the Genus Name field in the search screen. It is not necessary to enter anything in the Family Name field.

5. Periodicals. Until Supplement 16, volume numbers were cited in Roman numerals.

6. The publication date was not given for each record until Supplement IV.

7. In entries of new combinations, the author of the basionym did not appear in parentheses before the author of the new combination until Supplement IX.

8. Differences from the Printed Version. The printed version of Index Kewensis, up to Supplement 16, was converted into machine-readable form by filming the pages and then reading them with an optical character recognition scanner. Some characters could not be read by the machine; these were flagged and corrected later by hand. In addition, misreads affected about one per cent of characters. The optical reader also did not always format some records into the allocated fields correctly. These errors are being corrected as they are found and priority has been given to the family and genus names and the authority names. However, errors certainly remain, particularly in the Publication, Notes and Species fields. Progressive correction of these misreads (they generally centre around the confusion of similar characters such as 'b' and 'h', or 'c' and 'e') is planned and will continue over the next few years.

A further limitation of the optical character reader used at that time was that all accents and special characters had to be omitted. These might eventually be reinstated.

Other changes from the printed index include:

• Each record below the rank of genus now includes family and genus. In the printed version, all infrageneric records appeared under one generic heading.

• 'l.c.' (locus citatus), used when a number of names were published simultaneously in the same work, has been expanded so that all records now have a full reference attached.

• The dagger, used from Supplement 16 onwards to indicate the corrected re-entry of a record, now appears on the database as RE-ENTRY. Before the dagger came into use, re-entry was marked by a semicolon between two authors. The first mentioned is the author of the original entry and the second the author of the publication which follows the name. Sometimes this practice was used when the second author was validating a previously invalid name and at other times, because the second author was giving a more detailed description of the name even though it had been validly published by the first author.

• An asterisk, used in the printed version (Supplements X-XV) to indicate an illustration accompanying a description, appears on the database as ILLUS.

• The addenda to the original two volumes and to Supplements I and II were not read by the optical character reader. Many of the records in these addenda are incomplete, and some cannot be interpreted sensibly. New records with the prefix 'a' have been included in the database when the addendum entry could be understood.

Understanding the Gray Card Index data

Since computerization of the Gray Herbarium Card Index in 1992 efforts have been underway to standardize data in the Index and to bring it in line with internationally agreed standards. At present, approximately 98% of the data have been checked and standardized as follows:

1. Family names - Family names have been added to all the Gray Card records according to Index Nominum Genericorum. Some generic names may have incorrect family assignments because the same genus name may have at one time been applied to two different families, such as Calceolaria to both Violaceae and Scrophulariaceae, but most have been corrected. (100 per cent complete)

2. Genus names - All genus names have been checked against Index Nominum Genericorum and spellings have been standardized. For example, Aplopappus has been corrected to Haplopappus and Heleocharis has been corrected to Eleocharis. (100 percent complete)

3. Specific epithets - The spelling of all specific epithets is being brought into agreement with the ICBN. In doing so the ae in species epithets is being changed to i: for example, hypericaefolia to hypericifolia, absinthiaefolium to absinthiifolium, acaciaefolia to acaciifolia, etc. The original spelling is maintained in a comments field. (ca. 98 percent complete)

4. Author names - Author names are being standardized against Brummitt and Powell’s Authors of Plant Names (1992). Some names are unambiguous and have been easy to standardize - Linn. has been converted to L., Fern. to Fernald, T. & G. to Torr. & A. Gray, and H. & A. to Hook. & Arn. - but others have required careful checking in the original literature. For example, determining which ‘Gray’ is the author of a name: is it S. F. Gray, A. Gray, J. R. Gray, or one of 10 other Grays? (94 percent of authors names have been standardized)

5. Ranks - The names of all infraspecific ranks have been abbreviated consistently throughout: variety as var., subspecies as subsp..All ranks ever used in the Gray Card Index have been maintained, regardless of how peculiar or unfamiliar, including subsubvar., lusus, prolus, race, grex, etc. Searches for infraspecific epithets will return all names used below the name of species, regardless of rank. Following the recommendations of the ICBN, Greek designations of infraspecific categories, have been converted to the rank of variety unless specified otherwise in the protologue. (100 percent of the names of ranks have been standardized)

6. Publication titles - Publication titles are abbreviated according to Botanico-Periodicum-Huntianum, Suppl. (BPH/S) (1991) and Taxonomic Literature (TL2) (1976-1988) and its Supplements (TL2/S) (1992-2000) except in cases where TL2 gives the same abbreviation for different works. For example titles listed as Prodr. in TL2 are followed in parentheses by the name of the author. Example: Prodr. (DC.), Prodr. (Bauhin) and Prodr. (Swartz). (All but 76 publication titles have been standardized)

7. Volume number - All volume numbers have been converted to Arabic numerals.

8. Collectors - The names of all collectors are being checked against collector records where the collector’s name is cited by the initials of the given name(s) and the full last name*. The first name is spelled in full for collectors who would otherwise have the same abbreviation. For example, John J. Smith and James J. Smith are written in full, not as J. J. Smith, to distinguish them. The main, but not exclusive, source of information is the Index Herbariorum, Part 2 - Collectors, published by the International Association of Plant Taxonomists. (99.3 percent of the names of authors have been linked to authority files)

*We urge editors and authors to cite the names of all collectors by the initials of their given names plus their full family name. This will make the data in each record more precise and make searches by collector more accurate.

9. Country names - The names of countries have been standardized using the Times Atlas of the World (7th ed., 1985). In the original Gray Card Index the names of countries were cited in various ways over time. The Dominican Republic, for example, was cited in 22 different ways. Place names such as ‘Nova Granatensis’ have been converted to their modern country equivalents, Colombia in this case, but the original name is retained in a comment field. (100 percent of country names have been updated and standardized)

10. State/Province names - The names of states and provinces have been standardized using the Times Atlas of the World (7th ed., 1985) or Webster’s New Geographical Dictionary (1988). (100 percent of state/province names have been standardized)

11. County/equivalent names - The names of counties or their equivalents have been standardized using the Times Atlas of the World (7th ed., 1985) or Webster’s New Geographical Dictionary (1988). (100 percent of the county/equivalent names have been standardized).

12. Geographic Caution - Geographical information was not added originally, or precisely, for all names in the Gray Card Index. Some cards give only "Eastern Canada to Virginia", "Brazil to Argentina" or "British Columbia to Mexico". Such information is retained in a locality field. Where information on the original card applies to a single country, "Massachusetts to Virginia" or "Sonora to Coahuila" the name of the country has been added to aid in searching. The names of countries, states, province and counties are therefore lacking for some cards.

Back to IPNI Home