At this point, the problems begin:

(1) What is a State? The administrator does not know whether the term "State" includes the District of Columbia, the Commonwealth of Puerto Rico, Guam, the Virgin Islands, or the Commonwealth of the Northern Mariana Islands. All of the Department's other programs are in effect in D.C., and most of them are in effect in one or more territories.

(2) Which population figures? The administrator does not know which of several widely used measures of population he should use or how much discretion he has to select among data. For example, the Bureau of the Census, in Current Population Reports, estimates that the population of Alaska increased almost 4 percent between 1970 and 1976. But the 1970 census figure, based on a total survey, is more reliable than the samplings that show the increase. Should the administrator select the most recent year for which reliable data is available for all of the states, should he use the 1970 census, or should he use what he considers to be the most recent reliable data for each state even though this may mean using different years for different states?

(3) How to make payments? The administrator does not know how to pay the allotment to the states. Is it to be solely by reimbursement or may he pay in advance on the basis of estimates? In case you think this is an easy question, I invite your attention to 31 U.S.C. 529, "No advance of public money shall be made in any case unless authorized by the appropriation concerned or other law."

(4) What formula? Worst of all, the administrator does not know how to do the necessary arithmetic.

Even if the administrator solves the first three problems in a fashion that produces certainty and avoids litigation, he will discover that there are at least two plausible ways of performing the calculation for making state allotments, each of which produces different results. One way is to allot the entire $13 million among the states on the basis of population, and thereafter increase to $100,000 the allotment of each state that would otherwise fall below $100,000. In this case, he must select among alternative means of reducing the states that are initially above $100,000, in order to obtain the money to increase the other states. If the means he selects are simply to reduce all of those above$100,000 states pro rata, he will find that, in bringing up the below-$100,000 states, he has reduced one or more states below $100,000 that were previously at or above $100,000. He will then have to perform the computation a third time, and so on, until all states are at or above $100,000. Another possible way of making the allocation is to allot $100,000 to each state, so as to meet the requirement of a "floor", and thereafter to allot the balance among them on the basis of their respective populations.

Appendix B is a computer run that allots the $13 million in the two ways described. For several states the difference in allotments is large. The pro rata reduction method allots $1,036,300 to New York, compared to the initial-floor allocation method, which allots only $762,600: a loss to New York of $273,700. The initial-floor allocation method would cost California $309,300, but would increase the allotments for Puerto Rico by nearly 30 percent and Utah by almost 40 percent. If an ambiguous allotment formula were enacted, litigation would be a certainty because the amounts at issue would be enough to pay the states' litigation costs. Needless to say, your job as the draftsman is to identify the allotment method intended and draft it clearly.

...

Consider another example. The specifications at Appendix A tell us, "State requirements include . . the establishment of linkages between this program and law enforcement agencies and other agencies providing services to domestic violence victims . . ." As an administrator, what must you do to establish "linkages"? Do you make an occasional telephone call to other agencies or do you establish a highly structured arrangement among agencies for coordinated action? This specification reflects the policy officials' belief that a program to assist the victims of domestic violence would be more effective if it took account of other resources that the state can bring to bear. Unfortunately, it also reflects their unwillingness to subject their belief to serious analysis. If the bill is to demand linkages, it should either say what they are or provide a means for the Secretary and the state to agree on what they are. You will want to ask the policy officials for further guid

ance.

You have already observed the fog around the term "domestic violence". The term is a convenient way of expressing a complex and partly unformulated idea. This sort of compression causes a common problem in legislative specifications, which typically abound in verbal shorthand. Some of this jargon is indispensable. In the arcana of social security benefit law, for example, phrases used in specifications such as "dropout years", "guarantee cases", "old-starts", or the like, are convenient terms-of-art precisely alluding to complicated statutory provisions. You must know what the phrases mean if you are to draft social security amendments. On the other hand, the use of a term or catch-word, such as "domestic violence" or "linkages", may merely serve as a cloak for imprecision.

$1.6. Exercise in formulating clarifying questions. With experience, you will become sensitive to the parts of legislative specifications, such as those discussed, that are nebulous. As an exercise, write out

the questions that you will want answered in order to enable you to draft a bill from the specifications at Appendix A. When you have done this, compare your list with the one at Appendix C. In making this comparison, keep two things in mind. First, for

teaching purposes, Appendix C includes many questions that most experienced draftsmen would initially answer for themselves. Second, Appendix C excludes many questions that will later occur to the draftsman as he actually writes the bill.

$2.1. Objectives. The domestic violence specifications are for a free-standing bill. This chapter concerns itself with the principles that should govern the structure of such bills. A draft bill may, alternatively, seek only to amend existing statutes, or may contain both free-standing and amendatory provisions. Amendatory bills are discussed in chapter four.

To continue the focus on free-standing bills, assume that you have met with the policy officials responsible for the domestic violence specifications and that they have answered the Appendix C questions in the manner recorded at Appendix D. You are now ready to draft.

As your first step, you should divide your bill into bite-size chunks and, as a second step, arrange those chunks in some digestible way. Your aim is a framework that others can readily understand, remember, and retrace, and that future draftsmen can conveniently amend. Attaining these objectives for a draft bill calls for the exercise of intuition schooled by experience. Nevertheless, the ideas that followsome obvious, some elusive-may serve you as guides.

$2.2 Keeping your bill sections conceptually distinct. Statutes are divided into numbered sections. You can draw a useful suggestion on how best to make this division from a remark of the Canadian economist and author, Stephen Leacock, about his proficiency in languages. He observed that after only brief study of Latin and Greek he found that merely by glancing at a page of each he could tell which was which.

This brings us to the first rule of sectional division. It should be possible to tell which sections deal with which subjects. This may not strike you as an especially profound insight. Stated more pretentiously, it is the principle that each of your sections should be devoted to a topic that is conceptually distinct from the topic of any other section. This enables the reader to infer a section's content from its heading with assurance that the material for which he is searching is not also covered in other sections. To accomplish this you must first follow a coherent theory of division in allocating material among sections. Then, either you must draft your sections to be of the same order of generality or, if

the ideas of some sections are logically subordinate to those of other sections, you must reveal the logical hierarchy of those ideas in the sequence of the sections and in their captions. The ordering of sections is discussed at $2.6.

The Egg Products Inspection Act, outlined at Appendix E, shows a way to achieve conceptual distinctiveness. The Act's draftsman had to write requirements for the continuous inspection and sanitary operation of egg processing plants, and the condemnation of adulterated egg products. He grouped requirements for continuous inspection and condemnation in section 5 and placed the requirements for sanitary operation in section 6. The conceptual distinction between these sections is based on their theoretically different addressees. Section 5 instructs the Secretary when to inspect and condemn; section 6 instructs the plant operator to comply with sanitary regulations. In reality, the plant operator and the Secretary are interested in both provisions. But separating the sections in accordance with some logical principle, in this case the putatively different audience to which each is directed, allows the draftsman to ease the burden of locating and understanding the sections and to avoid overlap.

Conceptually distinct ideas do not necessarily call for separate sections. For example, the lawyer's analytical grid has separate compartments for (1) an instruction to inspect and (2) an instruction to seize for condemnation that which is inspected. Why did the draftsman combine these instructions in a single section, when it would have been so easy to illustrate their conceptual distinction by writing them as separate sections? Without being privy to the draftsman's thinking, you might nevertheless guess that his reasoning went something like this: The purpose of inspection is to assure quality by locating adulterated products: the consequence of locating an adulterated product is its seizure for condemnation. How better to underscore the interrelationship of these ideas than by including both of them in the same section?

Knowing when to separate ideas and when to combine them involves the balancing of intangibles in ways that no rules are likely to instruct. In the last case, for example, if joining the two concepts produced an interminable section, the draftsman might

have elected to separate them. Do not use different sections for different concepts, though, if the concepts are integral to each other, so that one of them, by standing alone in a section, implies the nonexistence of the other. In that case, the two concepts should either be in a single section or cross referenced.

A common violation of this last rule, the so-called "split amendment", is discussed in chapter four. The split amendment consists of two sections, one making an apparently unconditional amendment to a statute, and the other, as the reader discovers later in the amendatory legislation, causing the first section to be effective only for cases that are subject to some previously undisclosed contingency.

$2.3. Examples of sections that illustrate and sections that blur the relative subordination of their ideas. The ideal statutory structure is one in which each section deals comprehensively with a single topic, identified in its caption. It allows the reader to find within each section every rule that is logically subordinate to that topic. This means, of course, that every section is of the same generality—another way of saying that the subject matter of one section is not dealt with by another section.

Such an ideal statute should also contain only short sections because, other things equal, short sections are easier to read and understand than long sections. Unfortunately, the two principles-combine like ideas for logical coherence, but separate ideas for ready intelligibility-push the draftsman in opposite directions, sometimes with bizarre results. Title II of the Social Security Act has separate sections for its definitions of wages, employment, and self-employment. But there is also a section labeled "Other Definitions", which includes definitions of a wife, widow, divorced wife and divorce, child, husband, and widower (yes, in that order), plus a number of rules that do not look much like definitions at all (such as a subsection entitled, "When Periods of Limitation End on Nonwork Days", and one labeled, "Waiver of 9-Month Requirement for Widow, Stepchild, or Widower in Certain Death Cases, or in the Case of Remarriage to the Same Individual"). The principle of division is probably the relative length of the sections involved. Is that principle likely to help the reader locate a definition in title II?

In contrast, the Egg Act groups all of its definitions in a single section, section 4. If the definitions had been long and complex, as in title II of the Social Security Act, the draftsman might have assigned a separate section to each definition. An analog of this is the Internal Revenue Code's allowable deduc

tions from gross income. The 1939 Code collected these deductions in a single section, section 23. The user of the Code knew that there was only one section to which he needed to resort to study the rules that applied to a particular deduction. That section would, however, also include other deductions. In recognition of the growing complexity of tax law, the 1954 Code affords each deduction its own section. The user still need only resort to one section for a particular deduction, although now that section is more narrowly focused.

The clustering of definitions or deductions into a single section makes the section conceptually distinct from other sections. In the case of the Egg Act and the 1939 Code, definitions sections and deductions sections are not logically subordinate to the ideas that dominate the other statutory sections. Similarly none of the deduction sections in the 1954 Code, which devotes to each deduction an individual section, is logically subordinate to any other deduction section, because the basis for division is the difference in subject matter.

$2.4. An example of economy in drafting. A further goal in shaping sections is that of drafting economically. The Egg Act pursues it by using a single section to define many of its terms. Without any loss of clarity, these terms could have been defined each time they were used. By defining them only once for the entire Act the draftsman not only preserves our forests (an ecological objective rarely sought by legislative draftsmen), but avoids cluttering other sections of the bill with repetitious material. Later, we will consider this goal further.

$2.5 Exercise in subdividing specifications into sections. Before considering how the sections of a free-standing bill should be ordered, perform the following exercise. List every distinct idea or purpose in the domestic violence specifications at Appendix A (as clarified by Appendices C and D) that seems indivisible and worthy of expression by a separate section. Then place each specification under the section heading whose main idea seems to include it. No specification should be addressed by more than one section, and the sections, taken together, should cover all of the specifications. If the subject matter of a specification seems naturally to fall into two sections, you should either rethink your sections or divide the specification into two conceptually distinct parts.

Compare your effort with Appendix F. Keep in mind that when, as a later exercise, you arrange your sections in logical order, you will be free to aggregate bodies of related sections into separate

titles of the bill if this seems necessary to assist the reader in understanding the subordination of groups of provisions to more abstract concepts. At the planning stage of a bill, sectional divisions are tentative. As you draft the bill, itself, you may think of ways to improve on your original sectioning.

§2.6 Putting sections in the proper sequence. The sequence of the main provisions of the Egg Act, summarized at Appendix E, may be outlined as follows:

(1) Short title.

(2) Findings and purpose.

(3) Definitions.

(4) Principal operative provision, which also specifies the Act's scope.

(5) Subordinate operative provisions.

(6) Prohibited acts (and related exclusions from prohibited acts).

(7) Sanctions for commission of prohibited acts or other offenses.

(8) General administrative authorities and procedural rules.

(9) Jurisdiction of the courts.

(10) Relationship of the Act to other statutes.
(11) Administering agency's report to Congress.
(12) Appropriations authorization.

(13) Separability section.

(14) Effective dates.

This sequence is common in statutes that establish new programs. It enables one to understand a statute by reading its sections consecutively, as you would read the chapters of a book. Do not conclude from this that rules of composition that promote the clarity of draft bills are always those of an essayist. The draftsman is not employed to produce a work of literature, but to express legislative policy clearly and simply. Admittedly, definitions stacked at the beginning of a bill can be tedious and will delay the reader in coming to the "meat", the bill's operative provisions. On the other hand, this placement serves several important functions. It warns the reader early that certain terms have meanings that may differ from their dictionary definitions. Also, by immediately acquainting the reader in detail with the bill's subject matter, it makes the bill's operative provisions, read subsequently, more comprehensible.

As in the Egg Act, a bill's key operative provisions should come ahead of provisions having less scope. In short, the main material is up front. Administrative and technical provisions, or provisions of temporary effect (such as savings or grandfather provisions, repealers, and so on) come at the end. For example, the appropriations authorization (if any) for a regulatory statute, i.e., in most cases a

statute based on the commerce clause or the taxing power, is of limited interest and can be tucked away somewhere. In the case of a grant-in-aid statute, i.e., a statute based on the welfare clause, such as one that allocates appropriations among applicants, the appropriations authorization is of wide concern; for this reason, it is usually best placed near the beginning of the statute, preferably immediately ahead of the section that allocates the appropriations.

The Egg Act adopts an order that reveals the logical connections among that Act's sections and fits the reasonable expectations of the user. It is not an arrangement written in the heavens for all bills. If another sequence better serves the purposes for a specific subject, you should follow it.

§2.7. Exercise in sequencing sections. Using the principles described, place the sections of the domestic violence bill into an acceptable order. Then compare your work with Appendix G.

$2.8 Subdividing a section. A bill's sections are subdivided into subsections for the same reasons and with the same logic that the bill's subject matter was divided into sections. If a section's central theme is most readily understood when analyzed into its component subsidiary themes, you should divide the section into subsections. Each subsection should develop a single idea, readily distinguishable from, and ordinarily not logically subordinate to, each of the ideas upon which the section's other subsections are founded. One or more of those subsections may be further subdivided in the same fashion.

The draft bill at Appendix H illustrates some of the principles of subdivision. The bill is intended to govern the authority of the Secretaries of Agriculture and HEW to regulate the use of nitrites in food from its effective date. The specifications called for a bill that prohibited the Secretaries from banning nitrites before May 1, 1980, but allowed them to order nitrites off the market on or after that date, subject to certain procedural requirements. This temporal discontinuity divides section 3 from section 4. The principle of division was time, the principle of sequence was chronological order.

Section 4 would authorize action on nitrites after April 30, 1980, but had to be written to make the form of action depend upon which of two contingencies occurred. First, there was the possibility that nitrites might be shown to be safe. In that case, neither Secretary was to be permitted to ban them. Second, there was the possibility that nitrites might not be shown safe, but might be shown necessary to prevent botulism. In this second case, the Secretaries were to be allowed to ban nitrites after the lapse of a specified period. Separate subsections,