This paper presents an acoustic and auditory analysis of the short vowels of Mehri, a Modern South Arabian language (MSAL) spoken in Dhofar (southern Oman), eastern Yemen and parts of southern Saudi Arabia. Interest in Mehri vowels lies in the fact that phonologically distinct vowels are often phonetically extremely close: a fact of significance both for work on MSAL, with disagreement in the interpretation and transcription of vowels across various works (Rubin 2010), and for phonetic theories that claim vowels should be maximally dispersed (cf. Vaux & Samuels 2015). /a/ and /ə/, vowels with high phonological load, overlap across and within speakers, and the high long vowels /ī, ū/ are remarkably close to their corresponding mid-high long vowels /ē, ō/. Mehri has six long and seven short vowels. The long vowels are /iː, eː, ɛː, aː, uː, oː/ (Simeone-Senelle 1997). The full inventory of short vowels is /i, e, ɛ, a, u, o, ə/ (Watson et al 2020). There are, however, unusual features concerning the distribution of the short vowels. /u/ is only attested word-finally, /i/ is marginal, and /e, ɛ, o/ only occur in word-final stressed syllables, e.g. /ˈbkoh/ ‘to cry’, /imˈʃeh/ ‘yesterday’, /ɬ’iːˈɡɛʔ/ ‘caves’. The only short vowels not so restricted are /a/ and /ə/, forming a linear two-vowel (sub-)system. From a phonological point of view, this sub-system simply contrasts ‘high’ /ə/ with ‘low’ /a/; however, realizations not only range from front to back due to context, but also overlap considerably, as seen in the F1–F2 formant plot above. To show the range and overlap, we present acoustic data in the form of F1–F2 formant plots, accepted as the best acoustic fit to the perceptual vowel space (Kiefte et al 2013), and auditory data in the form of phonetic symbols placed in the IPA vowel quadrilateral, as below (blue only /ə/, purple /ə/ and /a/, red only /a/). Our research partly aims to see the extent to which our acoustic and auditory analyses tell the same story. Provisional results so far suggest that they do.