1.IntroductionPolymorphism is the occurrence of the same chemical compound in different crystalline forms. In addition, certain compounds are known to crystallise in two or more forms that are “similar”,i.e.the same 2-D or 1-D periodic fragments are assembled differently in each of these forms. Recent studies confirm that this phenomenon is not uncommon in organic crystals.1–5Among the examples are two forms of TNT4and the pigment Maize 1.5Furthermore, Gervaiset al.have reported on efforts to predict new polymorphic modifications by rearranging the potentially most stable periodic fragments of a known mother phase.6Assemblies of molecules which are similar may also occur in multiple-component crystals and crystals of close analogues. Such structure fragments are either 3-D (frameworks), 2-D (identical sheets, differently stacked), 1-D (identical rows, differently bundled) or 0-D (isolated units such as dimers).Any recurring periodic or discrete arrangement of molecules with its spatial characteristics may be called asupramolecular construct(SC).5One could argue that the frequent recurrence of a given SC in different crystal structure environments indicates its preference over alternative arrangements, whatever the reasons might be. It may then be concluded that the systematic search for SCs delivers implicit information on the assembly of molecules in the solid state. This idea lies at the centre of our strategy, and we disagree with the view7that “no special significance should be assigned to the similarity of crystal structures”.It is important to keep in mind that the concept of an SC differs from that of a supramolecular synthon,8employed in crystal engineering. The only condition for the existence of an SC is that a common spatial arrangement of molecules is present in two (or more than two) crystal structures. This concept is unbiased with regard to the nature of intermolecular interactions within these fragments of crystal structures. In contrast, the ‘supramolecular synthon’ focusses explicitly on how molecules are linked through definable interatomic interactions, primarily hydrogen bonds. Therefore, whilst a particular synthon could be in an SC, the same synthon can exist in fundamentally different spatial arrangements and SC's may assemble without involvement of identifiable synthons.Recently, we have decribed a method for the identification of SCs and the determination of the degree of similarity.5In our paper detailing these procedures, and the program XPac,5four case studies were presented in order to demonstrate the application of this method in different circumstances, using pair-wise comparisons of structures. In this article, we will demonstrate the use of the method for the systematic investigation of a large ‘family’ of crystal structures containing a particular component molecule. This is based on the CBZ molecule, with the family consisting of polymorphic modifications, multiple-component crystals and structures of close analogues. We aim to demonstrate the advantages offered by this new approach, in identifying relevant structural information that was previously inaccessible, in a reliable and efficient manner.Carbamazepine (CBZ,Scheme 1) is a pharmaceutically active substance with anticonvulsant and analgesic properties whose polymorphism is of considerable interest.9–13The structures of four polymorphic forms of CBZ have been characterised by X-ray crystallography. The high-temperature form1is triclinic and crystallises with four independent molecules.10Form2is trigonal.11The most stable polymorph at room temperature is the primitive monoclinic form3.12Polymorph4of CBZ isC-centered monoclinic.13All modifications contain N–H⋯O hydrogen-bondedantidimers in which thesyn-oriented NH group, but not theanti-oriented NH, is engaged in hydrogen bonding. Using graph set terminology,14the resulting rings are R22(8).Multiple-component crystals include co-crystals, hydrates, solvates, clathrates, inclusion compounds and salts. A number of multiple-component structures containing the CBZ molecule have been determined in a crystal engineering case study by Fleischmanet al.15The structures5–17(seeTable 1) produced by these authors fall into two categories. In5–10, the dimeric, hydrogen-bonded “homosynthon” observed in the polymorphs of CBZ is maintained, but the free H-bond donor site and the acceptor site of the CBZ molecule may be engaged in additional H-bonding with another crystal component. The same situation is found in the ammonium chloride (18) and bromide (19) complexes and the dihydrate (20) reported by Recket al.16,17Furthermore, the homosynthon is also present in both the furfural hemisolvate (21) and the dimethylformamidine solvate (23), reported by Johnstonet al.18,19Instead of the H-bonded dimers, hydrogen bonds between CBZ and carboxylic moieties are present in structures11–17. The hydrogen-bonded dimer is also absent in the CBZ/aspirin co-crystal22.20Details for structures1–25.Z′is the number of independent molecules of CBZ (CBZ analogue in24and25)Crystal SpGrZ′CCDCRef.Alternative descriptions of the same structure with a disorder model: (a) SpGr.Abam,Z′= 1, CCDC FEFNOT,ref. 17.SpGr.Cmca,Z′= 0.5, CCDC FEFNOT01.1CBZ form IP1&cmb.macr;4CBMZPN11102CBZ form IIR3&cmb.macr;1CBMZPN03113CBZ form IIIP21/c1CBMZPN01124CBZ form IVC2/c1CBMZPN12135CBZ/acetone (1 ∶ 1)P1&cmb.macr;1CRBMZA0115a6CBZ/dimethylsulfoxide (1 ∶ 1)P1&cmb.macr;1UNEYIV15a7CBZ/benzoquinone (1 ∶ 0.5)P21/c1UNEYOB15a8CBZ/terephthalaldehyde (1 ∶ 0.5)C2/c1UNEYUH15a9CBZ/saccharine (1 ∶ 1)P1&cmb.macr;1UNEZAO15a10CBZ/nicotinamide (1 ∶ 1)P21/n1UNEZES15a11CBZ/acetic acid (1 ∶ 1)P21/c1UNEZIW15a12CBZ/formic acid (1 ∶ 1)P21/c1UNEZOC15a13CBZ/butyric acid (1 ∶ 1)P1&cmb.macr;1UNEZUI15a14CBZ/trimesic acid/water (1 ∶ 1 ∶ 1)C2/c1UNIBAU15a15CBZ/5-nitroisophthalic acid/methanol (1 ∶ 1 ∶ 1)C2/c1UNIBEY15a16CBZ/adamantane-1,3,5,7-tetracarboxylic acid (1 ∶ 0.5)C2/c1UNIBIC15a17CBZ/formamidine (1 ∶ 1)P1&cmb.macr;2UNIBOI15a18CBZ/ammonium chlorideP21/n1VUBCAW1619CBZ/ammonium bromideP21/n1VUBCEA1620CBZ/water (1 ∶ 2)P21/c1This work21CBZ/furfural (1 ∶ 0.5)P21/n2FOMXAH1822CBZ/aspirin (1 ∶ 1)P1&cmb.macr;12023CBZ/dimethylformamidine (1 ∶ 1)P1&cmb.macr;11024CBZ analogueR3&cmb.macr;1BOWWAL2225CBZ analogueP21/c1VACTAU0123Furthermore, a fragment search of the Cambridge Structural Database (CSD)21identified two close analogues,24and25, of CBZ (Scheme 1).22,23These are of interest because the fundamental shape of the CBZ molecule is retained in both cases. The structure of25provides the only example of a N–H⋯O hydrogen bonded catemer synthon in this study. Details for1–25are listed inTable 1.First we will give a description of the methods employed to identify similar packing in the CBZ family,1–25. Thereafter, we will focus on the results obtained by this procedure and discuss the SCs and their impication.