The first step to be taken in the discussion is to enquire, What is the real normal death-rate of lying-in women? And, having ascertained this to the extent which existing data may enable us to do, we must compare this death-rate with the rates occurring in establishments into which parturition cases are received in numbers. We have then to classify the causes of death, so far as we can, from the data, with the view of ascertaining whether any particular cause of death predominates in lying-in institutions; and, if so, why so? And finally, seeing that everybody must be born, that every birth in civilised countries is as a rule attended by somebody, and ought to be by a skilled attendant; since, therefore, the attendance upon lying-in women is the widest practice in the world, and these attendants should be trained; we must decide the great question as to whether a training school for midwifery nurses can be safely conducted in any building receiving a number of parturition cases, or whether such nurses must be only trained at the bedside in the patient’s own home, with far more difficulty and far less chance of success. _MIDWIFERY STATISTICS._ It must be admitted, at the very outset of this enquiry, that midwifery statistics are in an unsatisfactory condition. To say the least of it, there has been as much discussion regarding mortality and its causes among lying-in women as there has been regarding the mortality due to hospitals. Yet there appears to have been no uniform system of record of deaths, or of the causes of death, in many institutions, and no common agreement as to the period after delivery within which deaths should be counted as due to the puerperal condition. Many of the most important institutions in Europe merely record the deaths occurring during the period women are in hospital, and they appear not unfrequently to do this without any reference to the causes. Similar defects are obvious enough in the records of home deliveries; and hence it follows that the mass of statistics which have been accumulated regarding home and hospital deliveries, admit of comparison only in one element, namely, the total deaths to total deliveries, and this only approximately. Dr. Matthews Duncan, in his recent work on the ‘Mortality of Childbed and Maternity Hospitals,’ has dwelt forcibly on these defects in midwifery statistics, and has made out a strong case for improvement in records. But, as will be afterwards shown, with all their defects, midwifery statistics point to one truth; namely, that there is a large amount of preventible mortality in midwifery practice, and that, as a general rule, the mortality is far, far greater in lying-in hospitals than among women lying-in at home.