This paper addresses four previously under-explored facets related to early childhood malnutrition, as manifested in stunting. First, it provides a new explanation for the \Indian Enigma": why Indian children are shorter than those from poorer countries. Our analysis, based on data for over 213,000 children, finds that the Indian height deficit is driven by the shorter height-for-age of children from the lower-ranked and stigmatised castes. Upper-caste children are taller than the average for low middle income countries, but stunting in the lower-ranked caste groups is 10-15 percentage points higher. This is true even after controlling for variables used in earlier explanations in the literature, viz., birth order and disease environment. Second, using repeated cross sections of national level data for India, we show that between 1998 and 2016 caste gaps in childhood stunting did not decline. Third, using longitudinal data from the Young Lives Project, we estimate the probabilities of transition in and out of stunting at various ages, conditional on being stunted at age 1. We show that children stunted at age 1 are 22% points more likely to be more likely to be stunted at age 15. Finally, we show that stunting at age 1 year significantly lowers cognitive and learning outcomes at ages 5, 8, 12 and 15 years. We explore the role of the enduring stigmatisation of lower-castes as the reason for the persistence of caste gaps. Our results underscore the importance of focusing on social identity and discrimination as key markers of malnutrition, and indicate that the origins of adult life disparities between caste groups lie in early childhood.