B-tree data pointer stored in the nodes in each layer, the data B + tree is stored in the leaf node, that when looking for a B + tree efficiency than B-tree logically it should be higher?
Under such circumstances, scattered data storage B-tree, and when to find on disk, you can not use the principle of locality, but the efficiency is lower.
B + node between the leaves as well as put together a list, if the query is a range, then you only need to find a front and a rear, intermediate traversing the list it
B-tree also kept to traverse the entire tree, in order to carry out the scope of inquiry, it is also slow.