Catalogs or Collections are administrative lists with inconsistent relationships to physical items. Therefore, a Cataloged Item is an abstraction, i.e., it is an item that has been cataloged, and hence defined, by the administrator of a catalog.
The term “specimen” is used synonymously with “cataloged item” throughout Arctos.
Cataloged_Item . Cat_Num VARCHAR2(40) not null
Catalog Number is the identifier assigned to a Cataloged Item. It must be unique (case-insensitive) within a particular catalog. Arctos currently supports three formats of catalog number.
- Integer: Catalog numbers must be positive integers. Arctos provides tools to predict the next number (e.g., data entry can leave catalog number blank), find gaps in sequences, search by ranges, clone full records, and provide GUIDs of the “expected” format. Sort may be numeric (1,2,,3…20). Only this format comes with these tools and abilities, and it is often worthwhile to convert existing mixed-string catalog “series” to Integer catalog numbers and store the old numbers as other IDs.
- Prefix-Integer-Suffix: An integer is required, and may be accompanied by a prefix and/or suffix. The concatenation of prefix-integer-suffix must be unique. No predictive tools are available. Search is by the concatenation or substrings thereof. Sort will be character (1,2,…20…3).
- String: Any unique non-NULL string is accepted. No predictive tools are available. Search is by the string or substrings thereof. Sort will be character (1,2,…20…3). Characters “/” (forward slash) and “ “ (space) are currently prohibited.
Cataloged Item Type
Cataloged_Item . Cataloged_Item_Type VARCHAR2(20) not null
A code table is available to explicitly label various types of cataloged material.
Coll_Object_Remark . Coll_Object_Remarks VARCHAR2(4000) null
Use remarks to document non-standard information pertaining to the specimen. Do not use remarks for any information which could be recorded with more structure elsewhere, including remarks better stored with a part, event, or any other “piece of the specimen.”
Coll_Object . Entered_Person_ID NUMBER(22) not null
Agent creating the catalog record in Arctos.
Coll_Object . Coll_Object_Entered_Date NUMBER(7) not null
Date on which the record was created.
Coll_Object . Last_Edited_Person_ID NUMBER(22) null
Agent last editing the catalog record.
Coll_Object . Last_Edit_Date NUMBER(7) null
Date on which the record was last edited.
Coll_Object . Flags VARCHAR2(20) null
Flags mark a specimen as missing information during the entry process. It is sometimes more convenient to bulkload data after the specimen record exists than to enter data with the specimen; flags serves as a marker to facilitate easily locating those specimens.
Coll_Object_Remark . Associated_Species VARCHAR2(4000) null
Free-text description of species associated with the specimen.
Collection . GUID_Prefix VARCHAR2(20) not null
Catalogs are most usefully defined by way of a GUID Prefix, which in conjunction with catalog number forms a unique identifier within Arctos, and in conjunction with Arctos’ URI forms a Globally Unique Identifier (GUID) for the specimen record. GUIDs, once formed, must never be allowed to change or expire. See deleting specimens for guidelines. All specimen citations should occur by way of GUID. Note that while GUID Prefix generally appears to be a concatenation of institution and collection code, it is in fact an independent concept; several collections from an institution may use the ‘Herb’ collection_cde (e.g. for vascular plants, cryptogams, and marine algae collections, for example).
Choosing a new GUID_PREFIX
We require GUID_PREFIX to be 20 or fewer characters, and contain exactly one colon
: not at the beginning or end of the string.
We recommend choosing a GUID_PREFIX that will survive being transferred through various mechanisms, displayed in web pages that might use “fancy” formatting, or typed by users with many kinds of input devices in many languages. We recommend only upper and lower-case ASCII letters (A-Z,a-z), other than the required colon.
Collection . Collection VARCHAR2(50) not null
A short name for a particular collection type. For example:
- Mammal Specimens
Collection . Collection_Cde VARCHAR2(5) not null
Collection Code links catalogs to collection-type-specific code tables. Values are generally descriptive (“Mamm” or “Bird”), but no deeper meaning exists. Existing collection_cde values are in a code table. Exploring the “filter” option of Attribute Type will provide an idea of how a collection type has been used.
Collection . Descr VARCHAR2(4000) null
An extended name/description of the collection. For example:
- University of Alaska Museum, Mammal Collection
- Parasite Collection at the Museum of Southwestern Biology, Albuquerque, NM
- Kenelm W. Philip lepidoptera collection
Collection . Institution_Acronym VARCHAR2(20) null
Abbreviation of the institution that hosts the catalog. For example, “MVZ” for Museum of Vertebrate Zoology, “UAM” for University of Alaska Museum (of the North), “MSB” for Museum of Southwestern Biology. Generally, these values are the same as those traditionally used for specimen citations within discipline-specific publications. There is now a demand for globally unique specimen identifiers on the Internet, and at least one registry for such abbreviations has been established.
From SpecimenSearch, Catalog Number accepts arguments of several forms. The following table is illustrative.
||No-operator inputs are string matched.|
||Dash-separated smaller–>larger integers specify a range. Note that there is a 1000-item limit on ranges and lists.|
||”=” (equals) prefix overrides all other operators and assumptions; only a matching string is returned.|
||“Second” item is smaller than “first” item; not considered as range.|
||“Second” item is zero-padded so not considered an integer; not considered as range.|
||Commas are treated as list delimiters unless the value is prefixed with an equals sign. Note that there is a 1000-item limit on ranges and lists.|
||Commas are treated as list delimiters unless the value is prefixed with an equals sign. Neither catalog numbers nor list elements must be numeric. Note that there is a 1000-item limit on ranges and lists.|
||”%” is “match anything.” This matches anything CONTAINING 12.|
||”%” is “match anything.” This matches anything ENDING WITH 12.|
||“_” is “match any single character.”|
||“_” is “match any single character.”|
Locating Specimens by Identifier
Each specimen in Arctos receives (and is define by) a single catalog number, along with any number of identifying numbers, often referred to as “Other IDs.” There are several ways, each with their own limitations, to search these numbers. The data available for searching vary wildly based on what collectors have recorded and what collections have entered. Some exploration is often involved in finding a particular set of specimens.
Along with catalog numbers, Arctos provides the capacity to attach any number of identifiers of various types to specimens.
Other Identifiers, like catalog numbers, have three components: A prefix, an integer, and a suffix. Individual collections define how these components should be used, acceptable values, and how data are to be entered, and these decisions affect what sorts of queries are possible. It is often not possible to deduce these rules and practices – contact us if you need help.
To get Other ID search, click More Options on the Identifiers pane of SpecimenSearch.
This will provide options to select Other ID Type and to provide an Other ID Number. (We generally use “number” in the sense of a license plate rather than an integer.) Additionally, you can choose whether the number is an exact match or a “contains” match. Exact match searches are case-sensitive.
It’s often unclear what type of ID might have been assigned to a number, and the descriptions currently do little to clarify that problem. It is therefore possible (and often most practical) to search by the number component, entirely ignoring ID Type.
The above example finds all specimens with any type of identifier (except catalog number)
containing the string
123. As of this writing, that search returns
9330 specimens. Additional criteria, coupled with Arctos’ sorting
capability, is hopefully enough to find the specimen data of interest.
To get all search options, click Customize (near “Show More Options”), select a “My Other Identifier” (which will also then appear in results and on various forms), and choose “Show 3-part ID Search.”
Click Close and the form will reload with total of eight search options.
For this example, we’ll use Collector
The simplest use case is to search for a string, here
This sends the query
display_value is a concatenation of prefix, number, and
suffix). This returns specimens with Collector Numbers of:
regardless of how the data were entered and are stored. (“ABC-1234-X” could be entered as prefix=”ABC-1234-X” or as prefix=”ABC-“, number=”1234″, suffix=”-X”; “1234” could have been entered as a number or as a prefix.)
Changing the dropdown from “contains” to “is” will, of the above examples, return only “1234.”
The “in list” option accepts a comma-separated list of values.
The above example sends SQL
(‘A’,’B’,’C’), and as of this writing returns three specimens:
The in range option works only for enforced-integer types of identifiers (currently only AF and NK). Attempting to use it for collector number will result in a datatype mismatch and return an error.
Three-part search to the rescue! (At
least in the cases where data are entered correctly.) All of the above
deal with the concatenation of prefix, number, and suffix. It is also
possible to search these independently. Search for integer
to send SQL
customIdentifier.other_id_number = 1234.
This is a numeric match of the numeric part of other IDs. It will not find specimens which have the numeric information entered into prefix. This information is not available to public users, but is evident from the edit form. This specimen will NOT be found with the previous search!
Prefix and suffix work similarly. This search:
AND upper(customIdentifier.other_id_prefix) LIKE ‘%A%’ AND
customIdentifier.other_id_number = 123 (note prefix is a
match and is not case-sensitive) and returns these specimens:
Understanding Cataloged Items
We address assigning catalog numbers to material with a few brief examples.
In short, we strongly recommend cataloging the item of scientific interest: the material that Researcher #2 will ask to borrow for confirmation when they find your citations in GenBank or publications. Any other approach complicates tracking citations and data management.
We present as example a brief list of things that may be cataloged in Arctos.
- A biological individual
- Standard practice in vertebrate collections, and the method we strongly encourage when possible. Biological individuals are generally the item of scientific interest, and the thing a future researcher will wish to examine if attempting to replicate results.
- A biological individual and their parasites
- Common practice in vertebrate collections, but makes locating or citing a parasite more complicated and less reliable than it needs to be. Rather, we recommend cataloging the host, cataloging the individual parasites (or donating them to someone who can), and establishing proper relationships.
- A lot (e.g., all intestinal parasites from an individual; all
members of a taxon from a time and place, or all insects from
- While lots are a convenient and sometimes necessary ”working group,” (e.g., due to the number of individuals involved or the available expertise in identification) we strongly discourage making lots available for citation. Insect collections often loan lots, and the borrowing researcher will sort the lot to individuals for which they are provided catalog numbers, a situation we find acceptable. Attaching cryptic and fragile “individual tags” to members of a lot when someone uses a specimen for molecular analysis makes little sense to us.
- An occurrence (e.g., each instance of the capture of
- This situation inevitably leads to confusing citations and bad science when an individual sampled multiple times at multiple locations is assumed by users to be multiple distinct individuals. Arctos supports cataloging encounters as specimen events under one cataloged item.
- Your “share” of an individual (e.g., tissues; the bones being
- Similar to occurrences in that this leads to multiple identifiers being assigned to an individual (and potentially the two being compared in a study), this should be avoided when possible. When unavoidable, both systems should support resolvable identifiers and link to each other, and specimen downloads should include the relationship. Arctos also adds a distinctive style to “same individual as” specimens.
- Various parts of an individual (e.g., tissues cataloged separately
- This denormalization of data inevitable leads to divergence and confusion (not to mention increased curatorial workload), in addition to the aforementioned implications of assigning the item of scientific interest multiple primary identifiers. Having reconciled the data in similar systems, we cannot possibly be vigorous enough in discouraging the continuation of such methodology.
- An entire collection
- We include this to stress the fact that cataloged items are wholly arbitrary concepts assigned to whatever someone wanted to catalog. That is, the scientific value of a cataloged number is entirely up to the person deciding upon the material to catalog.
- Several of the above
- An individual or physical item (or anything else) may have any number of catalog numbers within or across collections. While this is occasionally necessary for various political or administrative reasons, we strongly encourage avoidance, and the proper use of resolvable OtherIDs (in a system which supports them) to clearly link all of the components of the item of scientific interest together when multiple numbers are for some reason necessary. (see http://mailman.yale.edu/pipermail/nhcoll-l/2016-March/009178.html)
Understanding Specimen Records
As a highly normalized system, there is no real meaning to the term “the specimen record” in Arctos. Most views of the data provide information somewhat equivalent to Simple DarwinCore. No view contains everything that might be considered the entire specimen record. The available information varies wildly across records.
The minimum possible specimen record is a catalog number, although “core” data such as Identifications and Accessions are difficult or impossible to avoid in the interfaces. There are “we don’t know” values for all “required” fields and concepts; “unidentifiable” is a valid taxon which may be used in identifications, for example.
The “full specimen record” consists of the core data and all data linked to those data at any depth. The full specimen record should not be viewed as something that resides entirely within Arctos. There are currently no views which could be considered “the full specimen record.” A specimen record may contain any number of Attributes, Collectors, Citations, Specimen-Events, Identifications, Media, Identifiers, Parts, Transactions, etc. Many of these objects are linked to other objects, which are in turn linked to more. Some incomplete examples:
- Identifications may be linked to any number of taxa. Taxa may have any number of classifications, including those in external systems (such as WoRMS).
- Citations are linked to Publications. Publications may be linked to CrossRef, which may link to other publications (references and referenced by), author data in the ORCID system, funding data in FundRef, etc. Additionally, the context of other specimens - and therefore everything they are linked to, including more specimens - cited from publications may be critical to understanding the current specimen.
- Other Identifiers may form links to various sources of data including GenBank, UCMP’s Locality Database, specimens in other collections in and out of Arctos, etc.
- In addition to the direct link through Collectors, most “nodes” employ to Agents in various capacities - as publication authors, identifiers, attribute determiners, verifiers of specimen-events, etc. Agents in turn may contain:
- Any number of names
- Any number of status reports
- Any number of addresses, including any data which resides at dereferencable addresses (such as publications from ORCID, funding information through Projects, linked data from WikiData, etc.)
- Any number of Media, which may contain anything - images, text, videos, static or linked data, etc.
- The context of activity within Arctos, which often includes thousands of other specimens which may include everything mentioned here
- Any number of related Agents, including all similar information from them and the agents to whom they’re related
Any of these data - and those not mentioned here - may be critical to answering some questions involving a specimen, and should therefore be considered “the specimen record.” It is unlikely that any question requires all of these data, and assembling them into one view would be, at best, difficult. Arctos links to related data when possible, and we are always receptive to adding more or different data to “default views,” adding pathways to specimens through any of these data, or otherwise providing tools to access various data in the context of other data.
See also: Arctos as an ecosystem
Collections in Arctos are wholly administrative. Collections may be comprised of similar taxa (e.g. mammals), of various taxa organized for some purpose (such at the Hildebrandt Collection at MVZ), by legacy usage, or anything else. The sole functional or technical consideration is code tables, which are tied to collection type (collection_cde). For example, contrast Attributes available to a Mamm collection versus a Para collection.
Legacy collections often exist for various reasons, and these may have duplicate catalog numbers, unpredictable formats which may confuse users, or contain arbitrary divides which no longer make sense. Combining these into a unified collection in Arctos is generally trivial, and Arctos provides various mechanisms (such as actionable identifiers and redirects) to ensure that no functionality is lost. Collections with “less citable” catalog number schemes are unlikely to support actionable citations, and so little is lost if the “traditional catalog numbers” are subsumed under a “citable catalog number.” This approach has been used to unify and disambiguate several Arctos collections; we find tradition little excuse to go forward under systems which discourage good science.
Deleting records from Arctos
- Encumber the record(s) to be deleted. Create an appropriate encumbrance first, if necessary. Records may be flagged from individual specimens, or en masse by using the Manage widget from Specimen Results of a search. Once records are flagged, they may be deleted by users with the appropriate privileges.
- Find the encumbrance (under Tools). Click See Specimens and carefully review what you’re about to delete.
- From Manage Encumbrances, click Delete Encumbered Specimens. You’ll again be asked to review your decision, and must click the proceed button at the bottom of the page to delete the records from the database.
Note that there may be reasons to keep masked records in the database instead of deleting them.
It is sometimes necessary to move cataloged items from one collection or catalog number to another. When doing so, it is important to maintain a way of finding the specimen by its original identifiers. In this, be as specific as possible. Use specific identifier types and GUIDs if possible. (See more at Other IDs.)
Arctos provides HTTP redirect capability, under which one URL (http://arctos.database.museum/guid/KNWR:Ento:7193, for example) can be automatically redirected to another (http://arctos.database.museum/guid/UAM:Ento:228334). This helps in maintaining a record of the specimen rather than the specimen’s identifying numbers, and allows users to continue using bookmarks and links.
To do this,
- Ensure that the “old” URL returns a 404 HTTP status code. You may do
this in two ways:
- Delete the specimen. All users will then get the redirect.
- Encumber the specimen with a “mask record” encumbrance. Users who do not have rights to bypass the encumbrance (e.g., all public users) will then be redirected, while operators will be able to continue to access the record.
- Insert into table REDIRECT (Manage Data/Tools/Redirects) old and new paths. For example, if DGR:Mamm:123 is recataloged as MSB:Mamm:456, enter: old_path=/guid/DGR:Mamm:123; new_path=/guid/MSB:Mamm:456.
Other ID documentation has moved to it’s own page.