From time to time we receive communications from readers asking why we do not adopt the publication policies of the glorious 1920's. In those days, all papers were accepted without refereeing, page charge was unknown, and physics progressed at a tremendous rate with numerous important experimental and theoretical discoveries.We must realize, however, that physics research today is on a very much larger scale than it was thirty to forty years ago. In 1925 the Zeitschrift f\"ur Physik published 367 articles by 285 authors. In 1963 The Physical Review printed some 1600 papers by about 2500 different authors. The average length of the articles has more than doubled. On page 486 of this issue of Physical Review Letters, we find a communication which may be described as "twenty-seven authors in search of a plot."We believe that these numbers prove that the pursuit of physics has changed drastically in the last few decades, and the change is likely to continue. It is impossible to go back to the "good old days." On the contrary, we have been hanging on too long to procedures which were adequate in the past; we will soon need to make drastic revisions in our publication methods.One scheme which has frequently been proposed is that we institute a reprint service. Members would select a number of reprints instead of subscribing to the whole journal. Libraries would still receive the journal as it is at present. This system was tried by The Physical Society, London, for three years, 1953 through 1955. The cost of handling the individual reprint requests was too high. Perhaps modern automation methods could reduce this. But more significant is the fact that very little use was made of this service by the members of the Society. They simply could not make up their minds each month when they were asked to make selections. Those really interested had probably received preprints, or preferred to obtain reprints from the authors. It must also have been discouraging for authors to discover that so few colleagues were interested in their great work.It is obvious that the growth of physics demands an irreversible increase in specialization. Modern automation may have increased somewhat the amount of physics a researcher can produce, as well as the amount he can use. But the quantity he can read and learn has not changed. We need to modify our publication system to adapt it to the present circumstances. A number of innovations are now being studied with regard to feasibility, cost, and usefulness. We hope to come up soon with a solution to this problem.