We are seeing differences in the TYPES of crimes being committed over the last 100 years. For instance, in 1920 one would not expect a suicide bomber to walk into a small grocery store and attempt to massacre 30+ people by detonating explosive devices attached to their torso. However, just as theft was a serious problem in London in 1920, it still is today. For argument's sake, let us assume that there were more stabbings in London in 1920 vs. 2017. In fact, let us assume that the stabbings now are outnumbered 5 to 1 when compared to the number of stabbings in 1920. Even if this is true, we have a huge question that must be taken into consideration if we wish to determine whether crime rates are better or worse in 2017 vs. 1920: Is it the case that 20 stabbings per day in 1920 vs. 4 stabbings per day in 2017 indicates that crime rates are higher in 1920 vs. 2017 when we take into consideration the differences in the ways in which crimes are being committed? Perhaps there really are more stabbings each day, but even if there is one suicide bombing per month (this is just an arbitrary example to make a point), approximately 20-70 people could be killed in this bombing. Therefore, the death toll brought about by crime at the end of the month could be very similar if not even greater in 2017, but there are technically less crimes being committed because the suicide bombing is only one event.
These are questions we have to take into consideration if we wish to answer your question together @Morlaf
Disclaimer: I realize my numbers do not add up particularly well in the example used above (lol), but you get the point