M. Ishii $^{a}$, and A.M. Dziewo\'{n}ski $^{a}$
$^{a}$ Department of Earth \& Planetary Sciences, Harvard University, Cambridge, MA 02138, United States.
Although the presence of anisotropy in the inner core is generally accepted, its strength and depth dependence still vary considerably from model to model. Our joint inversions of normal-mode splitting-function coefficients and absolute and differential body-wave travel times have demonstrated that a significant part of the data can be explained by a depth-independent model with the symmetry axis aligned with the Earth's rotation axis. This model has weaker anisotropy compared to models based upon differential travel-time data and predicts almost a linear dependence of travel-time residuals on $\cos^2 \xi$, where $\xi$ is the angle the ray makes with the symmetry axis. The simple anisotropy model fits the PKIKP data trend well at all distance ranges except between $173^\circ$ and $180^\circ$ where the data exhibit strong curvature as a function of $\cos^2 \xi$. Because the data from this distance range are most sensitive to the deepest 300~km of the Earth, one may expect the deviation in data to be an artifact of poor sampling or bias. However, current data coverage is relatively good, and the curvature is consistently observed even if the data are divided into subsets, suggesting that it is a robust global feature unique to this distance range. Anisotropy of the central inner core is clearly different from that inferred for the upper part of the inner core: the inner-most inner core (IMIC) appears to be a seismically distinct region. Search for the location of the axis of symmetry for the IMIC suggest that it could either be at the pole or near the Sea of Japan. However current data set is not sufficient to determine a unique location. The existence of a small (corresponding to $\sim 0.01\%$ of the Earth's volume), distinct inner-most inner core has significant consequences. If the change in anisotropic behaviour is not pressure-induced, then the model restricts development of anisotropy to mechanisms acting close to the ICB. Mechanisms involving the entire inner core, such as degree~one convection, would preclude distinct IMIC anisotropy. It also suggests two distinct episodes of inner core evolution, presumably related to changes in core environment. For example, at an early stage of the Earth's history, the chemical composition of the core may have been different, or when the Earth differentiated, the core temperature and pressure were such that an inner core of a few hundred kilometer radius formed relatively rapidly. Alternatively, if the development of anisotropy is associated with geodynamo action within the outer core, the IMIC suggests a change in the magnetic field as the inner core grew.