With the first wave of the Baby Boom about to turn 70, firms are asking the hard question, earlier and earlier.

It’s the most famous retirement conversation in American legal history. In 1896 Justice Stephen Field, then in his thirty-third year on the U.S. Supreme Court, was slipping badly. His worried colleagues sent John Marshall Harlan to remind the 80-year-old that two decades earlier he had urged another failing justice to retire. As the story goes, Harlan found Field dozing in the robing room, woke him gently, and asked if he recalled the episode. Field looked at him with what has been described as “blazing” animus and burst out, “Yes! And a dirtier day’s work I never did in my life!” Harlan retreated and Field remained on the court for another year, setting a longevity mark that lasted until William O. Douglas’ tenure.

I was reminded of the Field story recently as I listened to a group of managing partners talk about their versions of the Dirty Conversation. These talks are prevalent now throughout Big Law. The pig in the python—the metaphor for the Baby Boom generation moving through American life—is coming to the end of its run. Next year, the oldest Boomers will turn 70, and behind them comes an age wave that will inundate farewell-party planners for the next decade.

What was striking about the discussion was the ubiquity of these conversations. At best they seemed to be regarded as necessary evils. When they went well, they led to graceful handoffs of clients and peaceful reductions in partner points. When they went badly they prompted Charlton Heston-style histrionics: I’ll give up my clients when you pry them from my cold dead hands! Despite the awkwardness there was an understanding that unless the firms repeal the iron laws of demography and aging, they had no choice but to pursue them. Firms might run from the Reaper, but they couldn’t hide.

What are the best practices in this area? There seemed to be five.

• First, firms need to understand their demographic situation. How old is the partnership? What’s the work history of each partner over the last five years? How much business do the Boomers still control? Nothing organizes the mind, one lawyer put it, like discovering that 60 percent of our business was controlled by partners 60 years old or older.

• Second, firms need to start these conversations early because they are difficult. For many lawyers, their work is their life and they may not want to go gently away. And even those who look forward to stopping can’t miss the intimations of mortality that are encroaching. (Of course, my friend, everyone needs an estate plan but you don’t really expect I’ll ever have to use it, do you?) By starting early with their partners, whether at age 55 or 60, firm leaders can eliminate some of the drama and make the end of a partnership seem a natural, expected transition. Of course by starting early some firms may spook colleagues to leave prematurely. But I assume that firm leaders already keep a list of flight risks, of those who can’t afford or bear to contemplate stopping: the 60-year-old who has just started a third family or the pale partner who has not taken a vacation since making partner in the first Reagan administration. The needs of those outliers can’t govern the firm’s policy, at least not at most places.

• Third, when it comes to clients, firms must not assume that they will blithely designate a successor. The general counsel of a Fortune 25 company grimaced the other week when he recounted to me how he’s weaning his company from his primary outside IP counsel. The lead partner is in his late 60s. He had made a few efforts to introduce younger colleagues but the GC didn’t think they were good enough to handle the work. So, slowly and quietly, he was diverting new engagements to other firms. His long-time counsel was secure for as long as he cared to practice, but when he closed his office laptop for the last time, the relationship with the client would be over. It turns out that clients can read a calendar too. And they never want to be taken for granted. Transfers of relationships, if they happen, take time and effort. And that process, like the rest of the relationship, needs to be managed and assessed. Another reason to start early.

• Fourth, firms need to strive for consistency even when exceptions are made to whatever policy is in place. This will contribute to a sense of fairness, which, leaving aside potential liability issues or bruised feelings, is something to which a fellow partner ought to be entitled. Some firms, typically lockstep partnerships, draw a clear line. No exceptions: you may keep an office but otherwise you’re done. Many firms offer leeway to managers who don’t care to push out older but still productive partners. Those firms need to be able to explain those decisions and treat them as precedent.

• Fifth, these exit strategies send a powerful message about the firm’s culture and expectations. At the discussion I attended, some firms proudly said that they did not have a firm wide policy. Instead they tried for meritocratic decisions based on each partner’s performance and continuing contribution. These tended to be smaller, often one-office shops, ones where name partners still walked the halls, though occasionally using canes. These firms were sending a message about how their partners could expect to be treated. So was the mega-firm that started the exit glide paths at age 63 and expected them to land by age 66. Their partners needed to get out of the way for those coming behind; it’s what their institution needed. They too were sending a message about their community and its expectations. Or, to put it differently, how firms help partners end their careers will speak loudly to younger partners about what the rest of their working lives may look like.

The good news in all this is that leaving a firm is not the moral equivalent of dying. With health, wealth, and a newly discovered taste for a little risk, lawyers can move to rewarding and valuable work. Firms can help. Ignoring the issue isn’t the preferred choice. Stephen Field was wrong.