History of the International Phonetic Alphabet

The International Phonetic Alphabet was created soon after the International Phonetic Association was established in the late 19th century. It was intended as an international system of phonetic transcription for oral languages, originally for pedagogical purposes. The Association was established in Paris in 1886 by French and British language teachers led by Paul Passy. The prototype of the alphabet appeared in. The Association based their alphabet upon the Romic alphabet of Henry Sweet, which in turn was based on the Phonotypic Alphabet of Isaac Pitman and the Palæotype of Alexander John Ellis.
The alphabet has undergone a number of revisions during its history, the most significant being the one put forth at the Kiel Convention in 1989. Changes to the alphabet are proposed and discussed in the Association's organ, Journal of the International Phonetic Association, previously known as Le Maître Phonétique and before that as The Phonetic Teacher, and then put to a vote by the Association's Council.
The extensions to the IPA for disordered speech were created in 1990, with a major revision in 2015.

Early alphabets

The International Phonetic Association was founded in Paris in 1886 under the name Dhi Fonètik Tîtcerz' Asóciécon, a development of L'Association phonétique des professeurs d'Anglais, to promote an international phonetic alphabet, designed primarily for English, French, and German, for use in schools to facilitate acquiring foreign pronunciation.
Originally the letters had different phonetic values from language to language. For example, English was transcribed with and French with.
As of May and November 1887, the alphabets were as follows:

1888 alphabet

In the August–September 1888 issue of its journal, the Phonetic Teachers' Association published a standardized alphabet intended for transcription of multiple languages, reflecting its members' consensus that only one set of alphabet ought to be used for all languages, along with a set of six principles:

There should be a separate sign for each distinctive sound; that is, for each sound which, being used instead of another, in the same language, can change the meaning of a word.
When any sound is found in several languages, the same sign should be used in all. This applies also to very similar shades of sound.
The alphabet should consist as much as possible of the ordinary letters of the roman alphabet; as few new letters as possible being used.
In assigning values to the roman letters, international usage should decide.
The new letters should be suggestive of the sounds they represent, by their resemblance to the old ones.
Diacritic marks should be avoided, being trying for the eyes and troublesome to write.

The principles would govern all future development of the alphabet, with the exception of #5 and in some cases #2, until they were revised drastically in 1989. #6 has also been loosened, as diacritics have been admitted for limited purposes.
The devised alphabet was as follows. The letters marked with an asterisk were "provisional shapes", which were meant to be replaced "when circumstances will allow".

1900 chart

During the 1890s, the alphabet was expanded to cover sounds of Arabic and other non-European languages which did not easily fit the Latin alphabet.
Throughout the first half of the 1900s, the Association published a series of booklets outlining the specifications of the alphabet in several languages, the first being a French edition published in 1900. In the book, the chart appeared as follows:
Initially, the charts were arranged with laryngeal sounds on the left and labial ones on the right, following the convention of Alexander Melville Bell's Visible Speech. Vowels and consonants were placed in a single chart, reflecting how sounds ranged in openness from stops to open vowels. The voiced velar fricative was represented by since 1895 until it was replaced by in 1900. too would be replaced by in 1931.
Not all letters, especially those in the fricatives row which included both fricatives in the modern sense and approximants, were self-explanatory and could only be discerned in the notes following the chart, which redefined letters using the orthographies of languages wherein the sounds they represent occur. For example:

the Arabic ain . is a simple bilabial fricative ... is the English hard th, Spanish z, Romaic θ, Icelandic þ; the English soft th, Icelandic ð, Romaic δ. is the non-rolled r of Southern British, and can also be used for the simple r of Spanish and Portuguese ... is found in German in ach;, in wagen, as often pronounced in the north of Germany . is the Arabic kh as in khalifa ; the Danish r; the Parisian r is intermediate between and. — and are the ha and he in Arabic. — and are sounds in Circassian .

Nasalized vowels were marked with a tilde:,, etc. It was noted that may be used for "any vowel of obscure and intermediate quality found in weak syllables". A long sound was distinguished by trailing. Stress may be marked by before the stressed syllable, as necessary, and the Swedish and Norwegian 'compound tone' with before the syllable.
A voiced sound was marked by and a voiceless one by. Retroflex consonants were marked by, as in. Arabic emphatic consonants were marked by :. Consonants accompanied by a glottal stop were marked by :. Tense and lax vowels were distinguished by acute and grave accents: naught, not. Non-syllabic vowels were marked by a breve, as in, and syllabic consonants by an acute below, as in. Following letters, stood for advanced tongue, for retracted tongue, for more open, for more close, for more rounded, and for more spread. It was also noted that a superscript letter may be used to indicate a tinge of that sound in the sound represented by the preceding letter, as in.
It was emphasized, however, that such details need not usually be repeated in transcription. The equivalent part of the 1904 English edition said:

t must remain a general principle to leave out everything self-evident, and everything that can be explained once for all. This allows us to dispense almost completely with the modifiers, and with a good many other signs, except in scientific works and in introductory explanations. We write English fill and French fil the same way ; yet the English vowel is 'wide' and the French 'narrow', and the English is formed much further back than the French. If we wanted to mark these differences, we should write English, French. But we need not do so: we know, once for all, that English short is always, and French always ; that English is always and French always.

1904 chart

In the 1904 Aim and Principles of the International Phonetic Association, the first of its kind in English, the chart appeared as:
In comparison to the 1900 chart, the glottal stop appeared as a dotless question mark at a slightly smaller size in the chart or sample transcriptions, similar to the footless form for the modifier letter, but as a full letter in the table showing script forms next to typographic forms. replaced. were removed from the chart and instead only mentioned as having "been suggested for a Circassian dental hiss and its voiced correspondent". is suggested for the Bantu labialized sibilant, and as a diacritic to mark click consonants. It is noted that some prefer iconic to, and that and are unsatisfactory letters.
Laryngeal consonants had also been moved around, reflecting little understanding about the mechanisms of laryngeal articulations at the time. and were defined as the Arabic ح and ع.
In the notes, the half-length mark is now mentioned, and it is noted that whispered sounds may be marked with a diacritical comma, as in. A syllabic consonant is now marked by a vertical bar, as in, rather than. It is noted, in this edition only, that "shifted vowels" may be indicated: for in-mixed or in-front, and for out-back.

1912 chart

Following 1904, sets of specifications in French appeared in 1905 and 1908, with little to no changes. In 1912, the second English booklet appeared. For the first time, labial sounds were shown on the left and laryngeal ones on the right:
was added for the Czech fricative trill and replaced, following their approval in 1909. Though not included in the chart, was mentioned as an optional letter for the labiodental nasal. was still designated as the "provisional" letter for the alveolar tap/flap. were defined as the Bantu sounds with "tongue position of θ, ð, combined with strong lip-rounding". were still included though not in the chart. was removed entirely.
For the first time, affricates, or ssibilated' consonant groups, i. e. groups in which the two elements are so closely connected that the whole might be treated as a single sound", were noted as able to be represented with a tie bar, as in. Palatalized consonants could be marked by a dot above the letter, as in, "suggesting the connexion with the sounds i and j".
were no longer mentioned.

1921 chart

The 1921 Écriture phonétique internationale introduced new letters, some of which were never to be seen in any other booklet:
replaced and replaced, both of which would not officially be approved until 1928. replaced and was added for a devoiced, but neither has appeared in any other IPA chart and the latter is not supported by Unicode. Also added were dedicated letters for the central vowels,, which appeared again in and in the chart in Le Maître Phonétique from 1926 to 1927, though without the Council's approval. Of these, only were approved in the 1928 revision, with a different value for, until were revived and regained the 1921 value in 1993. The old convention of was retained for where central vowels were not phonemically distinct. were still for obscure or indeterminate vowels, as opposed to the others, which would indicate clear pronunciations.
The book also mentioned letters "already commonly used in special works", some of which had long been part of the IPA but others which "have not yet been definitively adopted":

for a single-tap r
for the Czech fricative trill
for a voiced
for the Arabic ح and ع, "whose formation we do not yet agree on"
and for labialized sibilants found in South African languages
As "suggested":
* for Circassian dental fricatives
* for fricative of Bantu languages
* for a sound between and found in African languages and in Japanese
Small j for palatalized consonants:
Overlaying tilde for velarized and Arabic emphatic consonants:
for "dentalized palatals"
,,, etc. for retroflex consonants, previously represented by etc.
,,,,,, etc. for affricates
for the near-close equivalents of
for the near-open vowels in English not, man
for clicks, with for the common palatal click

It also introduced several new suprasegmental specifications:

for "half-accent"
for "reinforced accent"
Tones could be indicated either before the syllable or on the nuclear vowel: high rising, high level, high falling, low rising, low level, low falling, rise-fall, fall-rise
Medium tones, as necessary: mid rising, mid level, mid falling

It recommended the use of a circumflex for the Swedish grave accent, as in . It was mentioned that some authors prefer in place of. Aspiration was marked as and stronger aspiration as.
The click letters were conceived by Daniel Jones. In 1960, A. C. Gimson wrote to a colleague:

Paul Passy recognized the need for letters for the various clicks in the July–August 1914 number of Le Maître Phonétique and asked for suggestions. This number, however, was the last for some years because of the war. During this interval, Professor Daniel Jones himself invented the four letters, in consultation with Paul Passy and they were all four printed in the pamphlet L'Écriture Phonétique Internationale published in 1921. The letters were thus introduced in a somewhat unusual way, without the explicit consent of the whole Council of the Association. They were, however, generally accepted from then on, and, as you say, were used by Professor Doke in 1923. I have consulted Professor Jones in this matter, and he accepts responsibility for their invention, during the period of the First World War.

would be approved by the Council in 1928. would be included in all subsequent booklets, but not in the single-page charts. They would be replaced with the Lepsius/Bleek letters in the 1989 Kiel revision.
The 1921 book was the first in the series to mention the word phoneme.