Index, icon, symbol: a tale of abduction

Published in The Indexer 29:4 (December 2011)

In the semiotic theories of Charles Sanders Peirce, an index is a type of sign that signifies by having a direct connection to what it signifies – smoke is an index of fire, and a pointing finger is an index of what it indicates. The index is one of a trichotomy of sign types, the other two being the icon (which signifies by resemblance) and the symbol (which signifies by conventional association). Most semiotic constructions have elements of all three, and book indexes are no exception. The way signs are interpreted involves another trichotomy, of types of inference: abduction, deduction and induction. What readers take away from your index will depend on how you manage it – and your process of creating it – to optimize its indexicality, iconicity and symbolicity for optimal abduction.


I am about to commit the sin of equivocation: I will be examining indexes to see how much they are and are not indexes.

When you index ‘index’ in an index, of course, it matters which ‘index’ you’re indexing: digit, gnomon, table, librorum prohibitorum . . . . But I am evaluating what indexers make (indexes) in the light of the Peircean semiotic trichotomy of index, icon and symbol, as well as with reference to a related linguistic sense of ‘index’. And what will we take away from all this? Deductions. And inductions. And, perhaps most importantly, abductions.

The Peircean trichotomy

The first thing I need to do, before I precipitate trichotillomania, is explain what the Peircean trichotomy is all about. Charles Sanders Peirce (1839–1914), one of the foundational thinkers in semiotics (the study of signs and signification), determined that there are three kinds of signs – or, rather, three ways for something to signify, since most signs have elements of two or all three:

* An index signifies by direct connection – it is evidence of something, or it directs the attention to something.

* An icon signifies by resemblance – it has attributes in common with another object and so brings to mind that other object.

* A symbol signifies by convention – it is arbitrarily associated with the thing it signifies.

Peirce’s meaning of ‘index’ is similar, though not exactly identical, to the common usage of the term by linguists. Generally in linguistics, if something is ‘indexical’, it is drawing attention to something else; it may be evidence of it in the ‘where there’s smoke, there’s fire’ sense, or a direct pointer to it in the ‘Hey, look at that’ sense, or it may simply pull your attention and your thought processes to another thing that it belongs to or comes from, as, for instance, a particular accent is indexical of being from a particular place.

☞ So, for instance, if we look at a page and see a hand with a finger pointing at a paragraph (a typographical character called an index, as it happens), it serves as an icon in that it makes us think of a hand, and of a specific gesture at that (an indexical gesture: pointing); it serves as an index in that we know it proceeded from an intention on the part of the author, and that it is there on the page because of certain printing processes, and it is drawing attention to the paragraph by using a classic indexical gesture (or, rather, by being an icon of one); and it serves as a symbol because it has a codified form and existence in typography, and because, as a character used in a printed document, it surely has some particular conventional significance – why would one paragraph have one and not another?

But what if we actually don’t know exactly why it’s used there, and the author isn’t telling us? We make an inference, and Peirce came up with a trichotomy for that, too:

* Abduction looks at a phenomenon and forms a hypothesis to explain it on the basis of intuition or experience – it posits a rule (‘Oh, I see’).

* Deduction proceeds from premises to inescapable conclusion; we might say the premises are indexes of the conclusion.

* Induction starts with a theory – a hypothesis, and deductions on the basis of that hypothesis – and observes phenomena to see how closely they correspond with the theory.

☞ So, for instance, we might see the index (pointing hand) at the beginning of two different paragraphs, in each case followed by ‘So, for instance’, and we might make the abduction that it’s meant to draw attention to examples, or to treat them as digressive or secondary material. (We might also form hypotheses as to the author’s state of mind for doing so.) The deduction on the basis of that hypothesis would be that any paragraph giving an example is headed by the index symbol. Induction would involve seeing if that’s true. (Deduction and induction regarding the author’s state of mind may be more involved and are left to the reader as an exercise.)

There is one more trichotomy that we will be keeping in mind here, though it is not one of Peirce’s: the trichotomy of author, reader and indexer. Examples are not, I think, necessary here.

Creating and using an index

The question for us here now is the signifying functions and inferential processes involved in using an index – and in creating one. What of an index works because people discern and hypothesize a connection, what of it works because it falls into place as surely as Newtonian mechanics, and what of it works just because a convention has been set up and learned? What of the indexer’s work involves indexes – smoke from the fire (or embers) of the author’s mind; what of it involves icons – a thing that leads us to another thing by resemblance; what of it involves symbols – conventions that the author adheres to?

In order to determine how indexical an index is, we need to determine what functions truly are indexical, of course, and that’s not always as easy as it may sound. The following quotation from Peirce may or may not be helpful:

Indices may be distinguished from other signs, or representations, by three characteristic marks: first, that they have no significant resemblance to their objects; second, that they refer to individuals, single units, single collections of units, or single continua; third, that they direct the attention of their objects by blind compulsion. But it would be difficult, if not impossible, to instance an absolutely pure index, or to find any sign absolutely devoid of the indexical quality. Psychologically, the action of indices depends on association by contiguity, and not upon association by resemblance or upon intellectual operations. (Peirce, 1955: 108)

We could say that indexes are like a string attached to a thing. Guided by that image, we may take the existence of an entry in a book index, with headword and page number, as being indexical: it directs the attention to a specific place in the book as surely as a string attached to a pin attached to a map. We could equally say that each page number in an index is like a finger pointing.

But, on the other hand, an entry in an index does not occur by blind compulsion; it is not like the result of a computer search. Any indexer knows the amount of thought and choice that goes into the listings in an index. The association is also determined by conventions. An index does not as a rule say explicitly, for instance, ‘If you want to find out about bobcats, look at page 32.’ (Website links do, in fact, often do this, and so are more overtly indexical, especially the ones that say, for instance, ‘Click here to read more about bobcats.’) The word ‘Index’ in the title tells you that it’s there for looking up subjects, and when you see a headword you know that, at least in a well-indexed book, on the one hand it is not listing every instance of that headword (that would be a concordance, or the results of a search), and on the other hand on a given indexed page the headword itself might not appear as such, but the topic it names does. You know that the number after the headword means ‘look at this page’, and you know that if there are words indented below a word or listed in sequence after a word, they are subtopics to that word. All of that operates by convention – it may seem an obvious convention, and you may have determined some of it by abduction (‘Oh, I see’) the first time you saw an index, but it is convention no less.

These conventions can be changed without negating the indexical function of the index. Some books, for instance, index not by page but by section and subsection. Page numbers may or may not specify where on the page to look (encyclopedias sometimes use letters to indicate which of six or eight locations on the page to look at); they may also include other details, such as the name of the author in a multi-author volume. Some books have separate indexes for different kinds of information (linguistics texts often list example words in one index, persons referred to in another and topics in another, for instance), while others use typographical conventions (such as italics or small caps) to distinguish works, authors and topics, or pictures, diagrams and text. Some have multi-step indexes – Chinese dictionaries list radicals by number of strokes, numbering each radical, and then index full characters under their radicals. Some indexes add accessory information, such as colour keying to chapter tabs (which is both iconic and indexical); many include cross-references within the index; some use longer phrases as heads, while others use single words only. Entries are typically nouns or noun phrases, but a how-to book might have entries headed by verbs in the infinitive or imperative (‘devein shrimp’, ‘beat eggs’), a hymn book may have a first-line index, and a narrative might have an entry for a character name followed by subentries that are predicates for that subject (entry: ‘Jack Wilson’; subentries: ‘bakes a disastrous cake’, ‘offends party guests’, etc.). In some cases there will be some indication of the convention (such as titles on separate indexes, explanations of bolding versus italics); in others, it is left to the reader to see how it works – not just to say ‘Oh, I see how it works’ but also to understand that in this index they are to look for ‘eggs, how to beat’ rather than ‘beat eggs, how to’. In other words, not just the ground rules but their entailments for use are subject to abduction.

Nor should we forget that indexes are as a rule ordered entirely by convention. Nearly all are in alphabetical order, and the alphabet is an arbitrary ordering of an arbitrary symbolic system. Moreover, our handling of how to order items alphabetically is governed by further conventions: we don’t include articles (‘an’, ‘the’) in the criteria for alphabetization; persons are as a rule indexed by surname first (with numerous culture-specific requirements); in some cases, nouns are put before modifiers (‘chocolate, dark’; ‘cherries, Maraschino’); we have to decide whether ‘White Rock’ comes before or after ‘Whitehorse’. And then there are ‘Mc’, ‘Mac’ and their ilk. All of this is an arbitrary symbolic system, and yet there is often a choice of which variation on the arbitrary system to use – a choice that is nearly always left up to the reader to notice and say, ‘Oh, I see.’

The iconic index

But what is iconic about indexes? What in them signifies by resemblance? There are several things that might do so. For instance, we could argue that, in indexes with multiple levels of entry, it is iconic of importance to have main entries in larger or bolder type than sub-entries (largeness and importance are nearly universally equated). An index that ordered itself in the same order as the book’s chapters would be in an iconic ordering, but it would normally be called a table of contents. On the other hand, listing several references to the same topic in order of page number is, along with being a standard convention, iconic of their order in the book, while a listing in order of relevance (as search engines do) is iconic in that it shares the quality of prominence (iconicity can be a broad concept!).

More saliently, some indexes may use an icon of the subject: a text on anatomy, for instance, may have page numbers keyed to a diagram of the body, and a cookbook may have recipes listed according to their place on a menu (appetizer, salad, soup, main, side, dessert and so on). A work on geography might have a map with page numbers keyed to it (but of course maps generally have alphabetical indexes!), and a work on history might have a timeline with event names and page numbers. A child’s book about animals might have animals pictured (perhaps in order of size) and even buttons that allow you to hear the noise the animal makes; the pictures and noises would be iconic, even if their ordering were arbitrary.

The iconic approach to information access is more common on websites, where the reader may click on a diagram and then be taken to another page with a more detailed diagram or a listing of some sort (a site map). Site maps are iconic relationally, and in consequence may be thought of as tables of contents. Indexes on DVDs of movies (viewable through the Menu command) are iconic not only in being in the order of events but in showing stills, or even brief sequences, from the movie. And if you wanted, say, a character index for a DVD, it would be possible to have a screen of images of the characters, such that selecting one of them would lead to another screen with key scenes involving that character, perhaps with a brief emblematic action or statement that would play when the icon was highlighted.

Even within the restrictive means of alphabetical topic indexes in print, there is variable iconicity. The shapes of words have a certain iconicity, for instance, as the geniuses who came up with the clothing brand FCUK know very well (and the effectiveness of that iconicity is demonstrated by that word having leapt out at you as soon as you turned to this page); even the length of a word has a visual impact of its own. Capitals and numerals, due to their greater visual prominence, leap off the page (again, as demonstrated just above; this also makes digits much more useful than spelled-out numbers for reference purposes), and their common quality of size may be deemed iconic, although it does have an apparent indexicality of importance as well.

Choices of typeface (admittedly rarely under the control of the indexer) rely on resemblances within type styles and families as well as well-known usages within certain contexts; although it could seem trite for an index to use an Art Deco style of type for references to 1920s subjects, a digital calculator style for 1970s references, etcetera, it is an available option. Other elements are amenable to similar kinds of referential handling; just as restaurant menus say much about the style of restaurant depending on whether they list a price as, for instance, $18.50, or 18.5, or 18½, or ‘eighteen fifty’, so we can help set a tone for – or at least participate in – the style of a book with handling of elements such as page numbers. These are iconic just to the extent that they bring to mind qualities seen elsewhere; on the other hand, given that these features can be like speech accents, bespeaking a particular origin or attitude, they are also indexical.

Whatever style is chosen for an index, whatever ordering and presentation, they all have the character of indexes because they are all like sets of fingers pointing at individual instances, like strings on which you pull to produce that one single instance. The difference is in the choice of symbolic, iconic and indexical functions. The issue is really one of what will represent the author’s (or creators’) themes and concepts and characters best to the reader, and what will allow the reader to find what he or she wants to find in the way that he or she will find it most functional to find it.

Inductions and abductions

How does the reader know what the index is doing, or is meant to be doing? Often enough they know because they have seen similar things in the past, and at some point someone has explained to them how it goes. But there are acts of inference. The first viewing of an iconic arrangement (a map or diagram with page numbers marked) will involve the easy inference that the numbers are there for some purpose (that they are indexical of an intention on the author’s part), and specifically to indicate where in the book to find coverage of the location marked. Such an inference will be based on a general theory made from observations and tested over the years by induction. Something not seen before – perhaps an author’s name before a page number – will be handled by abduction: Why is it there? Why else would it be? Likewise, typographical stunts such as changing typefaces or colours according to the section of the book might seem strange and gratuitous at first, but through iconic relations (‘Oh, they use that type face in the chapter heading’) would be ‘figured out’. In all instances, there are some practices that a reader may not have encountered before, and is supposed to look at and say, ‘Oh I see.’ Creating the right environments for such abductions is a key part of an indexer’s work.

There is, of course, the risk of having a reader abduct something alien to the author’s and indexer’s intention. Page numbers keyed to a diagram might be read as indicating quantities; page numbers in a large work on history might be mistaken for dates in some contexts (especially if the context does not closely resemble a standard alphabetical index); numbers keyed to menu entries might be taken for calories, and so on. We must make careful use of our symbols and icons in order to avoid such alien abductions. Novel conventions are often best explained overtly, although reasonably plain ones may be left to the reader. There is, as well, a certain joy in discovering the association (when the light bulb goes on, you feel smarter). Websites often have home pages that require the user to discover what links to where by hovering over the link, and even in a book index small twists (as mentioned above) or cute stunts that take but a moment’s thought (perhaps a supplementary index to a biography consisting of a diagram of the subject’s body with page numbers keyed to brain, stomach, feet and other bits) might tickle the reader’s mind and increase enjoyment, which is a factor often overlooked in semiotics.

The indexer as high priest and medium

Does all of this make an indexer sound somehow like a high priest, with symbols and icons, or a heroic scientist fighting alien abductions? Well, an indexer is of course like a high priest – mediating between the human and the divine, or rather  the author and reader (which of those is the human and which is the divine is a question I will avoid answering) . The indexer is also like a heroic scientist, making hypotheses and deductions and testing them. The indexer must not only present icons, indexes and symbols, and engender abduction, deduction and induction; the indexer is also having to read the icons, indexes and symbols presented by the author and form abductions, deductions and inductions based on them.

The symbols presented by the author are generally the easiest part; the manuscript is made of words that are ordered in a certain way, and chapters are organized in this or that manner, and so on. All these things proceed according to an arbitrary code, a code that is learned and understood by all parties involved. (There are, of course, occasions when an author does things differently from how we are accustomed to see them, and the deviant code needs to be worked out.) The greater work of the indexer is in finding – in determining through observation and inference – the icons and indexes of the author’s structure and intention, of what topics and themes are important in the work, and of what the reader is going to want to find out, and how. Inclusion in an index is, after all, indexical of a degree of importance. The different levels of index entries indicate the relative importance and thematic relation of the items, and the indexer, in determining what to present and how, is not merely hierophant but haruspex, reading the entrails of the book to discern the patterns and foretell which the reader will seek where in the index. We might even say the indexer is a medium, communicating the spirit of the author to the reader – and vice-versa.

We might imagine that frequency of a word or phrase is a good index of its importance in a book. However, in some cases the name of a key idea or event is mentioned only a few times, while details of its explication or playing out (which may be second-level topics under it in the index) may be named many more times. In some cases a complete name is given only once, or even not entirely ever; in some cases, an index entry or subentry characterizes what the author is doing, although the term used never appears in the text as such (for example, under ‘semiotics’, the sub-entry ‘definition’). Other key terms may be used throughout, but may only merit indexing where they are explained (for instance, techniques and measures in a cookbook).

We could see the text of a book as terrain, and the indexer as mapping the peaks and valleys, plains and plateaux, lakes and rivers, aiming to name the peaks and rivers and to know what peak is a sub-peak of another peak, what stream a tributary of a river, what apparent plain actually an alpine meadow. The indexer abstracts the land of the book into an iconic representation, mentally (though perhaps not using the same icons or even being as consciously representative – intensity and density are as iconic as height and width), for the peaks and valleys are indexes of thematic importance just as the jags on a line graph are indexes of value. Thus an index is indeed like labels connected with strings to pins on a map, except that along with naming individual peaks the index also identifies points of commonality and difference between them.

And often the indexer’s preparations are iconic, laying out the flow; in all cases, they are indexical, marked at the location and also on a card or in a computer file. And yet they are deliberate choices by the indexer; they are not smoke from fire, but rather signs with arrows saying ‘fire’. In Peirce’s sense, they are indexical of the indexer’s thought processes, but in their function as references to key points in the book they are not operating by blind compulsion.

Can these preparations be more iconic? Some people like to make thematic diagrams or flow charts to aid their analysis of how to index, and in historical, geographic, anatomical, culinary or similarly mappable kinds of works, an actual icon of the subject matter is quite possible, even if only as an aid to the indexer and not intended for the final reader.

Can the preparations be made more symbolic? Additional conventions for representing different kinds of topic areas – which would arguably have an indexicality in the same sense as would a chosen style of dress for identifying someone as a part of a social in-group – are available, as are orderings in spreadsheets and similar applications.

And can the indexer’s work be more indexical in the Peircean sense? Computer programs and simple digital searches, which really do operate by blind compulsion, can make it so, but their limitations are well known and do not need further repetition here. But a good human-made index is an index (in Peirce’s sense) not so much of the work as of the state of mind of the indexer regarding the structure and salient points of the work, a state of mind which the indexer has worked to make iconic of the author’s and the reader’s to the extent possible, so that the reader can know which string to pull and can take away what is needed. And in order to do that, the indexer must also not simply pull the strings but discern the patterns and associations, and say, ‘Oh, I see!’


Peirce, C. S. (1955) Philosophical writings of Peirce, ed. J. Buchler. Mineola, N.Y.: Dover.

2 responses to “Index, icon, symbol: a tale of abduction

  1. Pingback: lepress | Sesquiotica

  2. Pingback: simulacrum | Sesquiotica

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s