One worker from Sliven, kept in Crematogaster schmidti at 0.898 — two thousandths under the table’s own bold line, the highest score in the file that is not printed bold, and until now only ever read as support for another body’s keeping
R1008 followed the stand-in at 0.945 and ended on the last type worker in the table with no page of its own. This page follows that worker alone. Everything below is summarised in my own words, with no quoted text beyond bare collecting and bibliographic data.
First, what the specimen is. It is a worker collected in August 1891 at Sliven, in Bulgaria — the revision’s distribution list gives Sliven at 42.68° N, 26.32° E, that month, leg. A. Forel — and it is now housed in Geneva, at the MHNG, under the identifier MHNG-ENTO-0296687. Its label, in paraphrase, names it as Forel’s variety christovichii of Crematogaster scutellaris — a spelling the revision itself flags with a [sic!] — from Sliven, Bulgaria, marked as a type. In 1892 Forel published the name properly as christowitchii, a variety of scutellaris; it was carried as a subspecies from 1893, and in 1922 Emery sank it as a junior synonym of schmidti. It shares all of that with exactly one other body: its fellow syntype from Pazardzhik, MHNG-ENTO-0296684, collected the same month. In the revision the two are almost never separated. The results passage speaks of them together, the account gives them one shared paragraph and one shared label transcription, and the sentence that affirms them — in paraphrase, the combination of their Bulgarian origin and the colour information in the original description, which described the animal in German as red with a black abdomen — affirms them as a pair. This worker has spent the whole revision as the second half of a plural.
Second, how it scored. Run as a wild card against the revision’s four clusters, formed from 201 workers in 68 nests, the Sliven worker came back at 0.898 for schmidti — with 0.088 for ariadnae, the new Cretan species, 0.014 for graeca and less than 0.001 for ionia. Every rival together holds barely a tenth of the reading. Set beside its fellow, the distance is wide: the Pazardzhik worker scored 0.620, so the two syntypes of one dead name stand 0.278 apart, one near the bottom of the table and one near its top. And yet, in Table 2, this worker’s 0.898 is printed in ordinary type. The table’s own footnote, in paraphrase, is a printing rule: values above 0.9 are set in bold. Three of the six type workers clear that line — the excluded Patras worker at 0.990, medispina at 0.984, the Dalmatian stand-in at 0.945. This one misses it by 0.002, the smallest possible-looking distance in the file, and so takes its place in the table’s visual grammar beside 0.711 and 0.620, among the numbers that look unbolded and therefore, at a glance, unsettled.
Third, what the bold line is — and is not. It is a fact about the table, not a fact about the ant. The assignment probability is a continuous quantity; nothing in the revision’s reasoning changes its character at 0.900. The proof is in what the revisers actually did on either side of the line. Below it, this worker and its fellow were not treated as doubtful: their synonymy is announced as confirmed by the morphometric analyses, their second-place reading for ariadnae is excluded on coloration and zoogeography, and the account’s affirming sentence is stated for the pair without qualification. Above it, the boldest number in the table, 0.990, did not keep its bearer at all — it expelled the Patras worker from ionia’s type series. Bold marks the strength of a measurement, never the direction of a verdict and never its security: a bold number threw a type out, an unbolded pair was confirmed where it lay. The two thousandths by which 0.898 misses the bold face are a typographic distance, one hundred and more times smaller than the 0.278 by which this worker clears its own fellow syntype. The animal did not come within two thousandths of anything. The printing did.
Fourth, what this worker has been used for. Until this page, its number has done its work inside another body’s story. In the account of the Pazardzhik worker — the lowest winning score of the six, kept down in schmidti only by its birthplace, a colour sentence from 1892, and company — the company is this: 0.898 standing in the same file is part of what lets a series be judged as a series, so that 0.620 never has to carry the synonymy alone. That is real work, and it is entirely outward-facing. Read as a subject instead of a witness, the Sliven worker turns out to need almost none of the scaffolding its fellow needed. Its runner-up is the same Cretan sister, ariadnae — excluded for it by the same zoogeography, since a worker pinned in inland Bulgaria in 1891 cannot be the Crete endemic — but at 0.088 that runner-up is a distant second, not the near-tie of 0.379 its fellow faced. Its own keeping, taken on its own, is the quietest in the file: a strong number, a forbidden rival, a birthplace that agrees, and a name that was already dead before any of the measuring began. The support was never supporting something fragile about itself. It was lending strength it did not need.
So what does 0.898 hold? A synonymy, jointly — the same dead name, christowitchii, held down in schmidti since 1922 and confirmed in 2026 — but it holds it from the strong end. The sequence of this file has now visited all six type workers in the table, and this is the only one whose story is a formatting accident: strong enough to be another specimen’s best evidence, never before read alone, and separated from the table’s visual mark of confidence by a distance too small to mean anything about an ant. A threshold drawn for readers sorted it, at a glance, with the doubtful. The revisers’ own text never did.
The walk — five moments in one worker’s working life, schematic only. Click each stage.
Pick a stage.
The excluded worker was certain and lost; the lowest worker was doubtful and carried; this one was strong and, at a glance, filed with the doubtful — because a line drawn for the reader’s eye ran two thousandths above its number. Read as the subject at last, it was never the weak witness. It was the witness.