It was based on the theory of Bell's Visible Speech, but set in roman script, and attempted to include the sounds conveyed by Lepsius's Standard Alphabet as well.

It differs from previous phonetic alphabets, especially the English Phonotypic Alphabet of the same author, by maximal use of trivial changes to existing characters, including rotated letters (such as ⟨ə⟩, ⟨ɔ⟩), small capitals (such as ⟨ɪ⟩), rotated small capitals, and italic rather than roman typeface (such as ⟨𝑙⟩).

⟨i y e œ æ ə a 𝑎 ɔ o u⟩ had basically the same values as in the IPA.

Not all of the vowels seem to make sense when plotted on a modern chart, as below, either through jumbled graphic correspondence or according to the languages they were identified with, suggesting that the phonetic analysis was not sophisticated.

A turned apostrophe, ⟨⸲⟩, essentially an ogonek, was used for nasal vowels, as in (a⸲) = IPA [ã], but ⟨ʌ⟩ for French nasal vowels (which phoneticists of this era described as having some sort of guttural quality), as in (aʌ) = IPA [ɑ̃].

(’h) was "a scarcely audible (ə)", and (’j) a "faint sound of (ᴊ, i)" as at the off-glide of the English vowel in see.

(lh) is a voiceless (l), but apparently not a lateral fricative, as Ellis renders Welsh 'll' as (lhh).

(‘b, ‘d, ‘g), defined as (b*p) etc., are unvoiced unaspirated (p, t, k) -- specifically the Germanic consonants frequently written [b̥, d̥, ɡ̊] in IPA.

Italic ⟨𝑡, 𝑑, 𝑠, 𝑧, 𝑡h, 𝑑h, 𝑛⟩ were used for the 'emphatic' sounds of Arabic and similar languages.

𝑝 might have been Lepsius's notation for the Quechua ejective p', in which case 𝑏 would have no explicit meaning.

E.g. (zs) for the final /z/ in English days, buzz, (sz) for German initial s. ⟨1⟩ (a turned digit 1) was used for the Semitic letter 'alif.

However, with other letters for alveolar consonants it was taken to mean 'advanced', with "the tip of the tongue on gums" (presumably dental, though that doesn't quite fit the second definition).

In several cases, Ellis carried over letters or at least the sounds that they described from Lepsius's alphabet without knowing what they were supposed to be.

(grh) was defined as (gh¿), a velar trill, which is not possible, and identified with the Semitic letter 'ayin, which elsewhere was given the value (↋).

The voiceless analog, (krh) = (kh¿), is identified with Swiss German 'ch' and Arabic 'kh' and so perhaps they were uvular (and duplicates of (ᴋh · ɢh)).or something between those values.

(fh, vh) are defined as (f*kh, v*gh), which do not occur in human languages.

(ɴh) was defined as the sound Lespius said occurred before Dravidian (ᴅh), rather than voiceless (ɴ) as one might expect.

He listed (ᴅhh), which he defined as "Lepsius's Dravidian sound, nearly (ᴅ𝑧h)", and so perhaps he intended ?