On the relationship of CSF pleocytosis to immunoglobulin levels as estimated by different techniques