The understanding of material self-assembling and self-organization mechanisms, at mesoscale, is crucial for nanotechnologies development. Such structures, hierarchically organized in superlattices or colloïdal crystals, are observed in inorganic or organo-metallic precipitates. The classical characterization, using electron microscopy and X-ray diffraction, of a synthesized powder is inadequate to describe the hierarchically organized polycrystalline structure of the final particles. Here, we have highlighted the successive steps that lead to the spontaneous formation of a cobalt oxalate dihydrate (COD) colloïdal crystals that precipitate in aqueous media. The characterization of the final particles, made by X-ray and electronic diffraction, has allowed us to determine the controversial crystalline phase of the precipitated COD. The β and γ COD phases has been discriminated by comparing experimental results with simulations of X-ray diffractogram and electron diffraction patterns for these both structures. The precipitated COD corresponds to the γ phase indicating that it comes from solid dehydration of cobalt oxalate tetrahydrate. The final powder characterization, performed by X-ray diffraction (XRD), atomic force microscopy (AFM), low voltage high resolution scanning electron microscopy (LVHRSEM) and transmission electron microscopy (TEM), gives some diverging results about monocrystalline or polycrystalline nature of the precipitates. XRD and AFM analysis show a polycrystalline structure of particles whereas LVHRSEM, TEM and electron diffraction observations indicate a monocrystalline structure. This apparent contradiction has led us to elaborate a new characterization method allowing to follow the growth and the structural evolution of the precipitate as a function of time. The cryo- preparation techniques are well adapted as they allow freezing and direct observation of solid state suspension by electron microscopy. The freeze/drying method has been modified and used for both SEM and TEM, and has allowed us to study the morphological and structural evolution of the COD precipitate from nanoscale to mesoscale. In addition, TEM analysis of samples prepared by classical techniques has been performed to complement cryo-electron microscopy observations. The combination of these two methods show that COD precipitation is a complex multi-step process leading to the formation of a core/shell heterostructure. The anisotropic core is porous and partially crystalline. It is formed by agglomeration of isolated primary particles and agregated ones (15-20 nm sized secondary particles). The crystalline shell corresponds to the final particles faces and is made up by the layer by layer self-assembling and alignment of 5-7 nm sized crystalline primary particles. These nanoparticles are aligned and perfectly ordered in strings of nanograins in the layer structure. The self-assembling occurs in the last step of growth where we have a lower ionic strength and supersaturation inducing slower kinetics of growth and aggregation. Moreover, the crystalline order in the primary and secondary particles increase with the time of reaction consistent with a continuous process of primary particle nucleation and growth. Our results show that self-assembling occurs layer by layer with terraces, kinks and steps that are present on the faces. This reminds the layer by layer crystalline growth models but in this case, the buiding blocks are colloïdal instead of atomic or molecular. Based on these different results, we propose an original model describing the COD precipitate growth. In addition, we have studied the poly(methacrylic acid-sodium salt) effect on the COD precipitation. This additive (PMMA) appeared to be a very good growth inhibitor. The PMMA effect on the final particles morphology is dramatic. Its concentration variation in solution slows down the nanoparticles aggregation rates along a  preferential direction. The crystal habit can then be controlled and we have synthesized platelets, cubes and rods with tailored length. The use of cryogenic electron microscopy has been shown to be a powerful tool for the understanding of time dependent aspects of particle growth. The use of such techniques for the study of other systems, where the final precipitate appearance may hide particle substructures, will help to elucidate growth mechanisms and allow finer control of nanostructured materials for many applications. The complex self-assembling and self-organization processes could then be highlighted and studied on a nanometer to micrometer scale.