From the first ensemble music I composed in the early 60s, I have used diagrammatic plans to portray in a proportional, rectilinear map, the prospects for the music I was setting out to create. Certainly, both my father's profession (architecture) and the engineering physics studies that I completed before undertaking formal musical training at the University of Michigan both influenced this. It did not stem from my teachers. Whether in formally prepared diagrams or informal sketching in notebooks, there is a geometrical cast to these beginnings. In early pages of the Angel project notebook, I am already assuming the desirability of a form that "allows for several presentations of identical or nearly identical materials/sections". The presence of the Sectional and Domain ideals is already known, and I ask myself "What about preparing a tape 'layer' that could be superimposed on either 1/2 of the 2- part overall structure, namely, whichever is last" This passage continues, "then the overall form would be either …" (cf. Ex. 9).
A few pages later, I am laying out an initial proposal for the definition of the Sectional part. It has a number of functions: Introduction, Theme, Transition, Combination, and "Other". There are, however, only three thematic elements at this stage.
Many of the features of the final plan are already in place. There is an introductory section that precedes the "Th 1". This initial theme is followed by a "TRANS 1" leading to "Th 2", which is, in turn, immediately followed by "Other". In this initial sketch, Other is surprisingly long, in proportional terms. In fact, it has an importance rather more as it actually turned out in the end (in contrast to what was proposed in the early overall plans [cf. Ex. 14 below]). Other does not have exclusivity, but coexists with a second presentation of Theme 1, and also with the beginning of a transition to Theme 3. This new transition also coexists with the first "Comb", which leads to Th 3. A second combination follows Theme 3, and appears to end the piece, although there is also a box below with unclear identity. The accompanying verbal list of elements includes a "closing" to balance the introduction, but this is not shown. If one replaces the introduction and closing posited here with thematic elements, the overall shape is closely related to later, more formal overall plans. The presence of two layers of narrow boxes in the present diagram is in keeping with the idea of stratified instrumentation (the piano contrasted with the ensemble), but, at this stage, I appear to be underestimating the potential dangers (technically, perceptually) of overlapping formal functions.
The second appearance of Theme 1 — now in the presence of Other — appealed to the psychologists, and survived in early versions of the overall plan, but was eventually discarded. Another feature of this earliest developed sketch is that, while Theme 1 occurs exclusively in the lower stratum, and Theme 2 in the upper one, Theme 3 is shared by both media. This feature was retained in the formal overall plans, but, in the end, was not accepted. In S, the fourth theme was presented by the ensemble only, and, in D, its core subsection was carried by the piano alone. Note that there is still only a general suggestion of proportion. The logarithmic series that shaped the proportions of the formal elements has not yet been evoked.
In the next notebook instantiation, more detail and precision is evident.
A listing of "TYPES of elements" at the right, includes 5 Themes, 3 Combinations, 2 Transitions, 1 Repeat, and 1 Other, and this is what the first formal overall plan included. Incidentally, one can note my realization that this number of musical segments adds up to 12 rather than to 11 (from the integer series). In the end, a repetition of Theme 2 under cover of Other was dropped and the number of formal elements reduced to 11. It became clear that the coexistence of formal elements — particularly because of the looming fact that the computer part would overlap either D or S in each performance — intensified the desirability of reducing the superimposition of ideas. An effort is made, at the bottom of Ex. 11, to resolve the proportions roughly indicated above into a more explicit form, corresponding with numbers from the proportional logarithmic series. The original "Introduction" has now become Theme 1, and other formal elements follow, more or less as before, though now the choice of specific durations for each has brought the alignment of the elements into a more refined state favoring exclusivity: one musical function followed by another. A further intensification of this tendency occurred as actual composition began, and, for example, Theme 3's appearance causes Combination 2/4 to become silent until the theme is finished. Similarly, Theme 5 replaces the indicated Combination of 1/2/3. (This combination will be discussed in more detail later.)
While the existence of a computer role in Angel has been mentioned up to this point, its actual placement has not yet been clarified. On a later notebook page, I reproduced a sketch that I had made in a meeting with my psychologist colleagues.3
It shows an S-D version of the overall form, indicating the presence of structural silences and also the fact that the computer was to enter after the first completed part, and to descend from a relatively high tessitura over the D part, ending in the deepest register. Though this strategic portrait remained relevant, it later developed that the computer would begin during the final moments of the first part (without an intervening silence which would have broken the form decisively), and that it would also end with a tumultuous summarization, after the live musicians had completed the second part.
At this stage, the need for the piano Epilog had not yet been recognized. Nor had the nature of the Domain version of the elements been fixed, nor the Sectional part been clarified in detail. The very rough sketch below proposes the nature of the two ideals.
At the top are three rectangular, "sectional" forms, with vertical slashes indicating the presence of sub-units. One rectangle is followed abruptly by another, displaced, without overlap. Below are three groups of concentric ovals, each with a small, central circle. The implication is that there will be a center common to both the S and D forms of the materials (i.e., a central or "core" element). In the Sectional traversal, musical functions are exclusive — one subdivided rectangle at a time — but in the Domain traversal, spheres of influence emanate from the core segments, strong nearby it, and weakening as they spread out. Thus, in the Domain version of the materials, rather than being sharply delimited, the influences of the various themes permeate one another, suggesting a more organic, constantly evolving spirit.
Finally, before embarking on a discussion of the compositional process, here is an advanced version of the formal overall plan.
In this version all relevant details regarding proportions are in place. The upper two strata represent the Sectional part. Each of the themes and also the transitions are given detailed subsectional definition. Theme 1, for example, is 61.5 seconds in length and has nine subsections of 3.5, 9, 5.5, 5.5, 14.5, 5.5, 3.5, 5.5, and 9 seconds duration. Transitions have, themselves, two independent strata within them, and these are given logarithmically-derived proportions. One can see that, according to sectional length, the influence of the initial theme (1) wanes by implication as its sections grow longer (15, 15, 24.5, and 40 second durations), while that of the destination Theme (3) grows more concentrated (28.2, 27.6, 14.1, 14.1, 13.3), and by implication, stronger. There is no inner definition yet for the Combinations, because, at this stage, I knew too little about the precise nature of the thematic elements to reliably predict how they would effectively interact.
he bottom two strata (the Domain version) are evidently related to the top two, but the lower picture is very different, indeed. For the first theme, in the upper layer, the 14.5-second (the fifth) subsection is the core. In the lower layer, the elongated rectangle beginning at the far left has an initial subsection length of 23.5 seconds, and then one of 14.5 seconds (directly under, and aligned with the core subsection of Theme 1 in the Sectional layer). The core of the Theme 1 Domain representation is followed by expanding subsections of 14.5, 28.5, and 56.5 seconds. A parallel circumstance can be traced for each of the themes: there is always a precise correspondence between the core versions of each theme in the Sectional strata and a corresponding subsection in the elongated, multi-part rectangles of the Domain representation below (cf. the plan's 23.5, 76, 208, 541, and 696.5 second marks). Each of the core themes is flanked by a group of subsections that grow in duration as they get further away from the core position. So, as has been explained above, the boundaries of influence for each thematic element in the sectional layer begin and end within concentrated durational limits of, progressively, 61.5 (T 1), 38 (T2), 23.5 (T 3), 23.5 (T 4), and 99.5 (T5) seconds. In the Domain layer, below, the parallel (now overlapping) timings for thematic statements are: 184.5 (T 1), 151.5 (T 2), 161 (T 3), 103 (T 4), and 131.75 (T 5) seconds. It is evident that the number of abrupt changes characteristic of the Sectional stratum is markedly reduced in the Domain version. And various other accommodations occur, primarily the fact that Transitions can be subsumed now into the overlapping influences of themes. Combinations are rendered symmetrical in the Domain layer to promote the sense of an emanating influence. Finally, the silences and the placement and size of "Other" are identical in both layers.4