Joseph Loschmidt: Structural formulae, 1861.

These diagrams are all taken from a booklet published in 1861 by Josef Loschmidt. The theme here is cyclic compounds, an idea that was still novel at the time. Kekule's famous proposal for cyclic benzene was still four years away! Loschmidt anticipates cyclic character for a diverse range of molecules, most of which have been subsequently discovered.

239: Diphenylchlorotriazene. Note how C=N is differentiated from triple bonded CN (structure 238).
Cyameluric acid. Discovered in 1850, this compound continued to baffle until 1937, when Pauling, on the basis of crystallographic evidence, proposed an entirely different structure from Loschmidt's imaginative meta-cyclophane structure here. Such m-cyclophanes were only made for the first time in 1969. A diagram for cyameluric derivative was left on Pauling's chalkboard when he died. Since no-one knew why he was interested in it, it was labelled "Pauling's mystery" and the chalkboard is now preserved for posterity. Cyameluric acid is mentioned once in his notebooks where the site of protonation is implied as OH (and not NH), but features prominently on his chalkboard, again with the wrong protonation!
Extended hydrogen bonding network predicted (PM5) to be formed by six cyameluric acid units.
344: Cyclic sulfate ester. Loschmidt draws S slightly larger than C (and inconsistently with O, which has a double circle). This compound is still known to this day as Carbyl sulfate, is the subject of several patents, and has exactly the same structure as Loschmidt drew.
330: Sulfuric acid. Why did he think that SO3 would dimerise (331)?
319: Thiirane (S is drawn as a single circle, rather larger than C)
Grappling with the Indigo/Isatin series. Note that in 278, the two central rings appear to imply five carbons. Something is not right here!
223 is benzoin, or 2-hydroxy-2-phenyl-acetophenone. Loschmidt draws a cyclic 1,3-dioxetane isomer! This compound remains unknown to this day, although the dimethyl derivative is implicated in various mechanisms.
This is where Loschmidt introduces the phenyl group, starting with 181, which shows alternating C=C and C-C allene structure, whih he discounts, following on with the layered sausage formula, which he also discounts, and ending with 185, which is meant to represent the uncertainty in the structure. Contrast this hesitation with its absence for 239. Had he substituted N in this latter structure by CH, he would have achieved the modern formula.
92 appears to be a cyclic anhydride. Again, this precise compound appears to be unknown to this day.
68: Cyclopropane
Cyclic N2O and linear N2O3. Not bad for pre Lewis days! Cyclic N2O was recently proposed (1999) as a reactive intermediate (Inorg. Chem., 38, 3006).
Diphenyl piperazine, a saturated six-membered ring

(C) H. S. Rzepa, May 6, 2005. Henry Rzepa is the 1996 recipient of the Joseph Loschmidit Prize for physical organic chemistry, awarded by the Royal Society of Chemistry.