A traditional "paper" edition of correspondence and other documents is surely nice and convenient to hold in one's hands, but is a rather rigid tool for study and research.
The Collected Interwar Papers and Correspondence of Roy Harrod, edited by Daniele Besomi and published by Edward Elgar in 2003, is surely no exception. The documents are there presented and arranged following the usual rules of textual editing: in particular, different kinds of documents are grouped in separate sections (volumes 1 and 2 contain the 1919-1939 correspondence, volume 3 Harrod's unpublished essays and his newspapers articles), in chronological order within each section.
Such an arrangement has its own logic: it enables the reader to see the flow of correspondence in and out in the chronological order, and can perceive how and to what extent the receipt of a letter or the outcome of a certain debate influenced the course of an author's thought (and possibly of the following correspondence ).
Yet there surely are other ways of arranging the materials, depending on a reader's specific interest. An obvious one is to read the correspondence between two authors, ignoring the other exchanges intercurring between letters. Some of these elementary needs can be responded to by using traditional means, such as cross-references in notes and headings and a rich subject index (see the Editorial introduction for the solution adopted in the Harrod edition). These primitive search and navigation tools work reasonably well, but are sometimes rather awkward to implement, e.g. when multi-volumes collections are involved or when multiple searches have to be carried out at once.
The search and navigation facilities on paper also suffer from another limit, the implications of which may be rather far reaching. All the cross-references and the entries in the subject index are prepared by the editor, and reflect his or her interpretation of the documents. To some extent, users are bound to the editor's mental schemes. If this takes advantage of the editor's "intimate and messy acquaintance"  with the documents, it may however hamper future research for instance by preventing readers to immediately perceive connections that may exist but have not been categorized in the index or have not been referred to in the editorial notes.
The editor's influence is also felt in the selection of the materials to be included in the paper edition, under the usual constraints of space and budget. The criterion for the selection of materials is usually that of their "relevance" or "importance", but this also reflects the editor's judgement.
A simple device like the html language, coupled with the possibility of search engines enabling to find strings or combinations of words, enables to remedy (partially at least) to these shortcomings. The navigation is much easier, as the "literary" cross-references are turned into active links which are, however, still supplied by the editor and are therefore subjected to the same limitations.
The search engine, however, introduces and additional degree of freedom: strings or combinations of words can be found in the entire collection of documents, regardless of their position. Such a tool should be used together with the subject index supplied by the editor, for not necessarily the same word or expression has the same meaning when used in different contexts, and concepts represented by different expressions are not captured by electronic (quick and thorough, but stupid) search engines.
On the web, space limits can easily be overcome: accordingly, the Harrod edition includes bonus materials (see the index), in the form of additional transcriptions and summaries, a gallery of portraits, and the reproduction as images of some originals. Moreover some secondary literature can be found on the web and is linked to whenever suitable, and the entire collection of image versions of the documents held in the Harrod and Keynes collection at Tokyo University can be linked to when appropriate.
On the other hand, some documents could not be included, either because copyright holders did not give permission (this is the case of the main repository of Harrod papers, that also owes the copyright of Harrod's writings in their possession) or because they asked exhorbitant fees.
This electronic edition, due to its features and limitations, is therefore not to be seen as an alternative to the paper edition, but as a complement to it, and it is offered in this spirit and also in the hope that such undertakings can, in the future, be routinely implemented.
 A striking example is the correspondence between Harrod and Kahn (beginning from letter 375) and between Harrod and Haberler (beginning from letter 364) in Autumn 1934: the latter exchange was deeply influenced by the former, as Kahn 'tutored' Harrod on the developments taking place in Keynesian though and Harrod incorporated them in his responses to Haberler. See for a discussion D. Besomi, " On the Spread of an Idea: The Strange Case of Mr. Harrod and the Multiplier ", History of Political Economy 32:2, Summer 2000. [return to text]
 Accordingly, the index in the Harrod edition explicitly indicates which entries refer to comments and indications supplied by the editor in footnotes or elsewhere in the editorial apparatus. Although this does not remedy the problem, it indicates that the problem exists and that the editor is aware of it. [return to text]
 The exopression is borrowed from Keynes's letter 791 to Harrod of 10 July 1938. [return to text]
 In the Harrod edition, this was addressed by indicating the selection criteria (see the Editorial introduction) and compiling a list of documents that could not be included in the selection (see the Personal and Routine documents). [return to text]
 For technical reasons, the subject index could not be included in the present edition (the links to page numbers did not turn into active links). [return to text]
 The corresponding pages are substituted with a warning to this effect. [return to text]